Planetary Rover Wheel and Lower Leg

Structural Design to Reduce Rock

Entanglements

Natalie Lawton

Space Engineering, master's level (120 credits)

2020

Luleå University of Technology

Department of Computer Science, Electrical and Space Engineering
 LuleåUniversity of Technology
Department of Computer Science, Electrical and Space Engineering

Planetary Rover Wheel and Lower Leg

Structural Design to Reduce Rock
Entanglements

Supervisor:
Dr.-Ing. Florian Cordes
Author:
Examiners:
Natalie Lawton
Prof. George Nikolapolous
Prof. Dr. Dr. h.c. Frank Kirchner

A thesis submitted for the degree of

MSc Spacecraft Design

13th February 2020

 Abstract
This thesis looks at the SherpaTT planetary rover. The rover is a hybrid walking and driving
rover that has been developed and built by DFKI and has already been deployed on several Mars
analogue field studies. The SherpaTT rover wheels were found to become entangled in rocks during
the last field deployment in Morocco. As human intervention would be impossible on Mars the
aim is to reduce the possibility of rock entanglements by performing a mechanical redesign of the
wheels. During this redesign care is taken to ensure the current traction, slip-resistance, weight
and strength are not adversely affected. In addition, the durability of the wheels is investigated in
terms of materials to review whether the current wheels are suitable for a mars deployment. An
investigation into the grousers design results in a changed design that aims to both reduce rock
entanglements and increase wheel performance by optimising the grouser height and number over
several different wheel and terrain cases. Wheels are produced for four scenarios, a rigid wheel on
hard ground, a rigid wheel on soft ground, a flexible wheel on hard ground and a flexible wheel on
soft ground. A conceptual investigation into the wheel fork design is carried out to examine the
effects of changing three properties of the wheel fork. The magnitude and location of the stress is
compared for each. Materials are investigated resulting in the recommendation of several potential
material choices which provide an increase in the overall strength and hardness. While SherpaTT
is still in development the 6000 class of aluminium is recommended due to the relative ease with
which it can be worked with. Once SherpaTT moves onto the final stages it is recommended that at
least the grousers are made from the 7000 class of aluminium, which have higher levels of strength
and hardness.
 Acknowledgements
I would first like to express my gratitude to my supervisor, Dr.-Ing. Florian Cordes, for all the
help and advice he has provided to enable me to complete this thesis and to Professor George
Nikolakopoulos for assessing my work.
I would also like to acknowledge my colleagues at DFKI, Bremen, who willingly gave their time to
provide feedback and observations on my work, thus facilitating improvements to it.
In particular I am grateful to Prof. Dr. Dr. h.c. Frank Kirchner, Andrej Kolesnikov, B.SC.
and Jonathan Babel, B.SC. for their invaluable time as well as the comments and criticism they
provided me.
I would like to thank all my friends for their support, especially Georges Labrèche, with whom I
have spent many hours locked in various stimulating discussions about Martian exploration.
Finally, I am appreciative of consistent support and encouragement during this sustained endeavour
that my family and my boyfriend have provided.
Contents

1 Introduction 8
1.1 Motivation and Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2 Rover System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Project Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Background 12
2.1 Parts Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Utah and Morocco Campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Past Work on the SherpaTT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Previous Mars Rovers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5 Previous Lunar Rovers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Current Work on Planetary Rovers . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.7 State of the Art Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.8 Martian Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.9 Potential Mission Locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.10 Wheel Terramechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.11 Durability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3 Design 37
3.1 Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Grouser and Wheel Fork Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Grouser Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Preliminary Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Durability Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4 Simulation Results 54
4.1 Flat Grouser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Chevron Grouser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3 Sawtooth Grouser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Rigid Wheel Spokes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.5 Flexible Wheel Spokes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.6 Flexible Wheel Rim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.7 Full Wheel Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.8 Comparison Wheel Fork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.9 Double Wheel Fork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.10 Taller Wheel Fork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.11 Slanted Wheel Fork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5 Final Results 75
5.1 Final CAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.2 Durability Investigation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

2
6 Discussion 81
6.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3 Improvements and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Bibliography 92

3
List of Figures

1.1 Sherpa Rover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.2 SherpaTT During the Morocco Field Tests . . . . . . . . . . . . . . . . . . . . . . . 10

2.1 Labelled Diagram of SherpaTT’s Wheel and Lower Leg . . . . . . . . . . . . . . . 12

2.2 Rock stuck on one of Curiosity’s six wheels . . . . . . . . . . . . . . . . . . . . . . 15
2.3 SherpaTT Wheel Fork Stuck on Rock During Morocco Field Tests . . . . . . . . . 16
2.4 Sojourner rocker and wheel assembly . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Two of the MER wheels with the yellow Solimide infill . . . . . . . . . . . . . . . . 17
2.6 Entrance and exit tracks of Opportunity after freeing itself from a sand dune known
as Purgatory Ripple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.7 One of Curiosity’s six wheels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.8 Wire mesh wheels from the Russian Lunokhod lunar rover . . . . . . . . . . . . . . 20
2.9 Wire mesh wheels from the American Apollo lunar rover . . . . . . . . . . . . . . . 20
2.10 Wire mesh wheels from the Chinese Yutu lunar rover . . . . . . . . . . . . . . . . . 21
2.11 Detailed view of Rosalind Franklin (Exomars) wheel . . . . . . . . . . . . . . . . . 22
2.12 Perseverance rover wheels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.13 Origami wheel fully extended and fully folded . . . . . . . . . . . . . . . . . . . . . 23
2.14 NASA’s most recent wire mesh wheel . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.15 Spherical rover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.16 Paddle wheels various configurations . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.17 Potential method of controlling a rover’s centre of gravity . . . . . . . . . . . . . . 25
2.18 Strain gauges interacting with wheel grousers . . . . . . . . . . . . . . . . . . . . . 26
2.19 Omnidirectional wheel potential configuration . . . . . . . . . . . . . . . . . . . . . 26
2.20 Map of Mars showing potential mission locations, previous succesful missions, previous
failed missions and planned missions . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.21 Nanedi Valles inner channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.22 Eastern termination of Naktong Valles . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.23 Mamers valley floor and one side of the valley wall . . . . . . . . . . . . . . . . . . 30
2.24 Valles Marineris topography and internal locations . . . . . . . . . . . . . . . . . . 31
2.25 Nilosyrtis Mensae layered deposits . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.26 Forces acting upon a static rigid wheel on soft ground . . . . . . . . . . . . . . . . 33
2.27 Scarecrow version of the Curiosity rover . . . . . . . . . . . . . . . . . . . . . . . . 35

3.1 A topographical map and other notable locales around Nilosyrtis Mensae . . . . . 38
3.2 A topographical map and other notable locales around Nilosyrtis Mensae . . . . . 39
3.3 Figure showing how the angle, spacing and height of the grousers can be altered. . 40
3.4 Figure showing how the shape and pattern of the grousers can be altered. . . . . . 40
3.5 Figure showing how the angle (b), clearance, thickness (f) and electronics (d/e)
could be altered on the wheel fork where (a) shows the original configuration. . . . 43
3.6 Graph showing how the rolling resistance varies with sinkage. . . . . . . . . . . . . 45
3.7 Graph showing how the grouser height varies with the grouser number and regolith
type. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.8 Grouser that uses a fold at each end to wrap around the edge of the wheel surface
where it is secured in place with a screw. . . . . . . . . . . . . . . . . . . . . . . . . 48
3.9 Grouser that uses a second piece of metal on the underside of the wheel surface that
pins it in place. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.10 Chevron shaped grouser attached via one long tab running along its length. . . . . 48

4
3.11 Grouser that has a zig-zag shape and is attached using metal tabs on either side of
the grouser. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.12 Grouser that has a wave pattern and small tabs to attach it to the wheel surface. . 49
3.13 The initial wheel designs using a chevron pattern a) and a straight pattern b). . . 50

4.1 Image showing the set-up of the flat grouser prior to simulation. . . . . . . . . . . 55
4.2 Figure showing the simulation result for the flat grouser under a load of 850 N. . . 55
4.3 Figure showing the set-up of the distributed load on the chevron grouser before
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.4 Figure showing the set-up of the concentrated load on the chevron grouser before
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5 Figure showing the simulation result for the chevron grouser under a distributed
load of 850 N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.6 Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring. . 58
4.7 Image showing the set-up of the sawtooth grouser prior to simulation where multiple
points are used to apply the force. . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.8 Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring. . 59
4.9 Image showing the set-up of the rigid wheel spokes prior to simulation. . . . . . . . 60
4.10 Figure showing the simulation result for the rigid wheel spokes under a load of 850 N. 61
4.11 Image showing the set-up of the flexible wheel spokes prior to simulation. . . . . . 62
4.12 Figure showing the simulation result for the flexible wheel spokes under a load of
850 N. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.13 Image showing the set-up of the flexible wheels inner rim prior to simulation. . . . 63
4.14 Figure showing the simulation result for the inner wheel rim of the flexible wheel
under a load of 850 N. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.15 Image showing the set-up of the rigid wheel prior to simulations . . . . . . . . . . . 64
4.16 Figure showing the simulation result for full rigid wheel design. . . . . . . . . . . . 65
4.17 Figure showing the simulation result for full rigid wheel design zoomed into the area
where the maximum stress of 6.692 ˆ 107 N {m2 is occurring. . . . . . . . . . . . . . 65
4.18 Image showing the set-up of the conceptual version of the comparison wheel fork
prior to simulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.19 Figure showing the simulation result for the conceptual model of the comparison
wheel fork with a maximum stress value of 3.219 ˆ 109 N {m2 . . . . . . . . . . . . . 67
4.20 Figure showing the simulation result for the conceptual model of the comparison
wheel fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2
to allow an easier comparison with the other models. . . . . . . . . . . . . . . . . . 67
4.21 Image showing the set-up of the conceptual version of a double wheel fork prior to
simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.22 Figure showing the simulation result for the conceptual model of a double wheel
fork with the scale showing the maximum stress of 1.452 ˆ 109 N {m2 . . . . . . . . 69
4.23 Figure showing the simulation result for the conceptual model of a double wheel fork
with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow an
easier comparison with the other models. . . . . . . . . . . . . . . . . . . . . . . . 69
4.24 Image showing the set-up of the conceptual version of the taller wheel fork prior to
simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.25 Figure showing the simulation result for the conceptual model of the taller wheel
fork with a maximum stress value of 3.994 ˆ 109 N {m2 . . . . . . . . . . . . . . . . . 71
4.26 Figure showing the simulation result for the conceptual model of the taller wheel
fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow
an easier comparison with the other models. . . . . . . . . . . . . . . . . . . . . . . 71
4.27 Image showing the set-up of the conceptual version of a slanted wheel fork prior to
simulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.28 Figure showing the simulation result for the conceptual model of the slanted wheel
fork with a maximum stress value of 3.058 ˆ 109 N {m2 . . . . . . . . . . . . . . . . . 73

5
4.29 Figure showing the simulation result for the conceptual model of the slanted wheel
fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow
an easier comparison with the other models. . . . . . . . . . . . . . . . . . . . . . . 73

5.1 Figure showing the final CAD for a rigid wheel on coarse sand. . . . . . . . . . . . 76
5.2 Figure showing the final CAD design for the rigid wheel on soft sand. . . . . . . . 77
5.3 Figure showing final CAD for the flexible wheel on hard ground. . . . . . . . . . . 78
5.4 Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring. . 79

6
List of Tables

2.1 Martian Regolith Types and Classification . . . . . . . . . . . . . . . . . . . . . . . 27

3.1 ESA Martian soil properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.2 Grouser equation results showing what the predicted optimum height and number
of grouser would be if the SherpaTT current grouser height or grouser number was
kept the same. The predicted values are shown in italics. . . . . . . . . . . . . . . . 47
3.3 Table showing suggested grouser parameters for four cases. . . . . . . . . . . . . . 47
3.4 Mechanical Properties of Several Considered Materials . . . . . . . . . . . . . . . . 51
3.5 Mechanical Properties of Metal Alloys which Remain Ductile at Cryogenic Temperatures 51

7
Chapter 1

Introduction

This chapter introduces the motivations, scope and goals which provide the basis for this thesis.
Additionally a brief introduction to the Martian rover system that the project is been based upon
is included.

1.1 Motivation and Scope

Since humankind first started to explore space the goal has always been to develop a deeper
understanding of the universe surrounding us. In order to accomplish this there are five potential
methods. The first is studying from afar using a telescope, the second is using an orbiter to fly
close to or around a particular celestial body of interest, the third, fourth and fifth are placing
a lander, a rover or human explorers on the surface of a celestial body. Which method is most
appropriate depends upon the type of celestial body or event, the distance from Earth and the
type of science that is being targeted.
For this thesis Mars has been chosen as the celestial body of interest. Mars today is a barren planet
with little atmosphere, however, there is evidence of a much wetter past [1], [2], [3]. By learning
more about Martian history scientists hope to gain a deeper understanding of both Mars and of
the planetary processes which have caused such changes in the Martian climate. It is additionally
hoped that humans can colonise Mars [4], [5] and searching for clues to Mars’ wetter past should
be able to provide ways in which humans can sustain themselves on Mars. Whilst there is an
aim to send humans to Mars this does not seem likely in the near future due to the huge costs
involved, the new technology that must be developed and tested to ensure humans can be sent and
returned safely from Mars and the increased risk of cross contamination between Earth and Mars.
Given this, there has been substantial interest in Mars with many countries and space agencies
sending a collection of orbiters, landers and rovers to the planet [6]. Orbiters have returned much
important information about Mars, including finding suitable landing sites for the landers and
rovers. However orbiters cannot complete any direct science on the Martian surface. For this a
lander or rover is required. Although a lander can return great surface data it is restricted to the
area in which it landed. As landing areas must be safe enough to land in they are often relatively
flat spaces, however, some of the more interesting science takes place down in the Martian valleys
and craters and up on the Martian mountains and volcanoes. For this a rover is required and in
order for a rover to move around efficiently wheels are required.
The wheels on a planetary rover are a critical component and have a huge impact on the mission
if they are to fail for any reason. Unlike Terrestrial rovers it is not possible to carry out repairs on
planetary rovers and as such any damage they take is permanent. In the best case scenario a failure
would lead to reduced mobility and in the worst case would completely immobilise the rover. If a
rover is completely immobilised it is no different from a lander. Therefore it is important that the
wheels do not fail.
There are several processes in which wheels could fail including;

8
 • Degradation of the wheel leading to the wheel falling apart.
• Failure of the wheel motors.
• Failure in an electrical connection between the wheel and the main power or logic boards.
• Entanglements in rocks or other surface debris inhibiting movement.
• Excessive sinkage in soft surfaces.
• Inadequate grip or traction on slippery or steep surfaces.
This thesis focuses on the rock entanglement failure mode and more specifically on rock entanglements
with the wheels grousers and the wheel fork. To enable this the SherpaTT rover will be used as
the basis for this research.

1.2 Rover System Overview

The SherpaTT rover is a robot designed and built by DFKI [7]. It is a hybrid walking and driving
rover which is able to transport modular payloads in order to carry out science. Due to the
hybridisation of the walking and driving modes SherpaTT is able to access areas that may be
otherwise unreachable by more traditional rovers whilst still maintaining an efficient drive motion
across more benign terrain. SherpaTT is the successor of the Sherpa rover, both rovers are briefly
described in the following sections.

1.2.1 Sherpa

The original Sherpa rover, seen in Figure 1.1 was constructed as part of the RIMRES project which
aimed to develop technologies which would allow complex tasks to be carried out in difficult areas
through the use of modular and re-configurable robots [8]. Using an actuated suspension system
the Sherpa rover was able to drive both with wheels and by using a stepping motion when the
terrain became too difficult to traverse purely by driving [9]. The Sherpa rover is now retired as
it’s successor, the SherpaTT, is complete.

Figure 1.1: Image showing the original Sherpa rover

9
1.2.2 SherpaTT

SherpaTT, seen in Figure 1.2, is the successor of Sherpa and builds upon the strengths of Sherpa
whilst improving on the Sherpa’s weaknesses [10]. SherpaTT maintains the hybrid driving and
stepping capabilities of the Sherpa rover, retaining the ability to traverse over more complex
terrain. Changes were made in the legs where knee joints were added providing an increased range
of movement, whilst two rarely used joints that provided the 6th Degree of Freedom (DOF) were
removed [11]. SherpaTT therefore has an increased degree of flexibility in movement compared to
Sherpa. SherpaTT has now completed two field tests, one in Utah and the second in Morocco,
where the rover has been tested in environments similar to ones it may encounter on Mars.

Figure 1.2: SherpaTT During the Morocco Field Tests

The biggest goal of the SherpaTT rover is to complete planetary exploration missions, with key
targets being Mars and the Moon. By using its high flexibility with its movement SherpaTT is
able to traverse terrain that is impossible for a rover that purely drives. This freedom of movement
allows for the exploration of previously unreachable areas. It also makes the rover more robust
than previous rovers in terms of entrapment as SherpaTT can walk out of many entrapments that
would permanently immobilise a rover with only a drive mechanism.
SherpaTT also has the potential to be used in Terrestrial applications where it is too dangerous
to send humans and the terrain too complex for a wheeled rover. Examples of this include areas
containing nuclear waste or underwater environments. A waterproof version of SherpaTT already
exists in the form of SherpaUW [12].

1.3 Project Aim

Whilst the SherpaTT is able to walk its way out of many difficult situations it has been noticed
in field trials that two types of rock entanglement have been occurring. The first entanglement
occurs when a rock becomes wedged between the wheel tread and the grouser. The rock then
travels with the wheel as it turns until it reaches the top of the wheel where the clearance is too
low to allow it to pass. When this happens the rock jams the wheel preventing it from spinning
any further. This entanglement disables the motion of the wheel affected. The second is when the
rover is passing over or close to larger rocks which then become entangled in the rover wheel forks.

10
When this occurs the rover beaches itself upon a rock leaving the wheel spinning freely in the air.
SherpaTT must then save itself by using the walking motion to step off the rock. Even though this
entanglement does not have as significant an effect as the grouser mode it is still an undesirable
occurrence for a rover that is intended for use in more challenging terrain. These entanglement
issues are covered in more depth in Section 2.2.
This project’s main aim is to make improvements to the current SherpaTT wheels and lower leg
structure to reduce the number of instances of rock entanglements. The primary ways in which
SherpaTT is currently becoming entangled is via the grousers and the wheel fork. Therefore, in
order to achieve reduced entanglement the grouser and wheel fork design are updated.
Whilst completing this primary aim the project also aims to ensure that other aspects of the
wheels performance are not adversely affected. Therefore careful consideration is made to the
grouser traction performance and the wheel fork stress profile.
Finally the durability of the wheel and lower leg structure is evaluated to ensure that it can
withstand the harsh Martian environment. Recommendations are made regarding materials and
structures.
The aims that are addressed by this thesis can therefore be summarised as follows;
• Reduce rock entanglements with the grousers
– Redesign grousers to remove primary methods of entanglement
– Investigate how the size, shape and number of grousers affects performance
– Rework wheel design to accommodate new grouser set-up
• Reduce rock entanglements with the wheel fork
– Investigate different wheel fork shapes
– Compare the magnitude and location of the stress in different wheel fork shapes
• Propose design to increase the durability of the wheel and lower leg structure
– Recommendations regarding wheel and grouser material
– Recommendations regarding wheel fork structure and material

1.4 Thesis Structure

This thesis is structured as follows:
Chapter 2: Background This chapter introduces the SherpaTT rover and the rock entanglement
problem in more detail. It then discusses past, current, future and state of the art work within
the field of rover wheels. The Martian environment is then explored and potential mission
sites identified. Wheel terramechanics is then introduced and the chapter ends with a note
on durability.
Chapter 3: Design This chapter takes the lessons learned from chapter 2 and begins to iterate
the design. The mission location is narrowed down and the key grouser and wheel fork
parameters are identified. Preliminary designs are found and a durability investigation takes
place.
Chapter 4: Simulation Results This chapter displays the how the simulations were set-up, the
results and the interpretations and conclusions for each of the designed parts.
Chapter 5: Final Results This chapter shows the final wheel designs and explains how they
were arrived at. Notes are made on the wheel fork structure that is best suited for the
wheels. The chapter ends with the results from the durability investigation.
Chapter 6: Discussion This chapter summarises the work undertaken in this thesis and presents
the final conclusions. It then talks about what future work could be completed leading on
from this project.

11
Chapter 2

Background

This chapter first provides a more detailed introduction to the SherpaTT rover, explaining the
different parts of the rover before talking about past field test campaigns and elaborating further
on the rock entanglement issues. Next the chapter describes previous and planned Martian rovers,
focusing on the wheel subsystem. Finally the Martian environment and wheel terramechanics are
covered.

2.1 Parts Introduction

The parts of the wheels and lower leg structure on the SherpaTT are listed and described here to
provide an overview of the wheel and lower leg structure system. Figure 2.3 shows where each part
is within the wheel and lower leg system.

Figure 2.1: Labelled Diagram of SherpaTT’s Wheel and Lower Leg.

12
2.1.1 Wheel Surface

The wheel surface, also known as the outer wheel rim, is the main body of the wheel that has
contact with the ground. The design of the wheel surface determines the shape, size and width of
the wheel. A connection to the rest of the lower leg structure is made through the attachment of
the wheel surface to the springs and spokes.
It is important that the wheel surface is durable enough to withstand driving over hard and
uneven surfaces without cracking or deforming beyond the point of failure for the wheel. As it is
important to minimise the weight of the rover where possible, providing enough durability is often
a delicate balance with the thickness of the wheel surface. Whilst wear on the wheels is inevitable,
particularly if the rover goes beyond its expected lifespan, there should not be any catastrophic
damage due to inadequate durability during a wheel’s expected lifetime. Additionally for rovers
with flexible wheel surfaces, such as SherpaTT [10] and the Exomars Rosalind Franklin rover [13]
the wheel thickness plays an important role in determining how much the wheel is able to flex, with
a thicker surface providing a more rigid wheel and a thinner surface a more flexible wheel.

2.1.2 Grousers

The grousers are attached to the surface of the wheel and are generally spaced at regular intervals.
They provide increased traction for the wheels by utilising increased shear forces when they make
contact with the ground and by increasing the wheels effective radius when the gaps between the
grousers are filled with soil. Research also indicates that having a higher number of grousers and
having taller grousers acts to increase the total drawbar pull, though the benefits become less
significant after a certain point [14], [15], [16]. Sharp edges on the grousers also allow the wheel to
bite into rocks and other hard surfaces when travelling over them thus decreasing the amount of
slip [17].
On the current SherpaTT model the grousers have a saw-tooth edge and are placed in pairs with
a 45˝ angle normal to the wheel surface. SherpaTT has 10 grouser pairs.

2.1.3 Springs

The springs in SherpaTT’s wheels connect the wheel surface to the spokes. The springs allow
the wheel to flex more than it would using stiff spokes. The current SherpaTT wheel deforms by
approximately 7mm under a load of 166kg, SherpaTT’s nominal mass [7]. Given that Hooke’s law
states that Fs “ k ¨ x, where F is the force acting on the spring, k is the spring stiffness and x is the
displacement of the spring from its original position, the spring stiffness of the springs currently
used on SherpaTT is therefore 233 N{mm.

2.1.4 Spokes

The spokes connect the springs to the wheel motor, sensor and wheel fork. They also provide
extra rigidity to the wheel and act to limit the maximum deformation of the wheel possible under
extreme cases.

2.1.5 Wheel Fork

The wheel fork connects the wheel to the main leg of the rover. This important component must
be able to withstand many different stresses having the weight of the rover above it exerting force
downwards, upward forces from the wheel moving over uneven terrain below it, turning torques
acting on it from the wheel steering and shear forces as the wheel traverses slopes.

13
2.1.6 Sensors

The wheel torque sensor is currently located in the central part of the wheel and lower part of the
wheel fork, measuring both the wheel spin torque and the forces acting on the wheel. Looking at
the torque measurements allows it to be easily seen whether the rover is moving or stuck when the
motor is running, while the force readings provide information on the ground contact of the wheel
which allows for active ground adaption control.
As this sensor currently sticks out of the wheel, increasing the width of the lower leg structure,
there is the potential to move it into the the upper part of the wheel fork where it would measure
the wheel direction torque instead.

2.1.7 Wheel Drive Motor

The drive motor is located in the centre of the wheel and provides the turning force to spin the
wheel. The drive motor dictates the minimum size the central part of the wheel can be which
impacts on the rest of the wheel design when the wheel size is limited by constraints such as
launcher size.

2.2 Utah and Morocco Campaigns

SherpaTT has currently been on two field campaigns, the first in Utah and the second in Morocco.
Due to the rockier nature of the Morocco campaign rock entanglement was not seen until this
campaign.
The Utah campaign took place in 2016 making use of the similarities of the Utah desert and Mars
to validate the SherpaTT’s ability to function in such an environment. Three different natural
terrain test tracks were identified for the rover to follow to observe the rovers performance in
several different scenarios. A flat surface of 20 m was used for the first, for the second a partially
flat 12 m track was used with a 4 m long slope of approximately 8˝ ’s inclination during the central
part of the traverse and finally a 17 m sloped track with up to 28˝ ’s inclination. All runs were
completed at a constant velocity of 0.1 m{s with the exception of the last runs on the sloped track
which experienced 100 % slip after the duricrust was broken and had to be completed at 0.4 m{s
[7].
The Morocco campaign took place in 2018 also making use of the similarities between the Terrestrial
environment in the Moroccan desert and Mars. However the Moroccan field test environment had
some differences to the Utah environment. In Morocco the rocks were far more numerous that
in Utah. The increased prevalence of rocks brought to light some previously unseen issues. It
was noticed that the rover was having several negative interactions with the rocks where the rover
would become entangled via two main situations.

2.2.1 Rock Entanglement via Grousers

This first rock entanglement situation was the more common of the two. Rocks would get stuck in
between the grousers and the wheel surface. They would then travel with the wheel as it turned
and become stuck in in the gap between the wheel and the top of the wheel fork. The wheel would
then be unable to turn freely immobilising the wheel.
During the design phase of SherpaTT this rock entanglement possibility was considered, however
it was thought that any rocks which became trapped in this method would be able to simply fall
off naturally without causing any problems to SherpaTT. As this has since proven to not be the
case this thesis will investigate the causes to the rock entanglement and propose design changes to
prevent or reduce entanglements.
There are several factors about the wheels current design that could be contributing to this
situation. First, these rock entanglements could be due to the angle at which the grousers are

14
mounted at. As they are mounted at 45˝ , as seen in Figure 2.1, to the wheels surface this makes
it easier for rocks to become wedged in this gap compared to grousers which are mounted at 90˝ .
Second, the clearance between the wheel and the top of the wheel fork is fairly small. Having a
small clearance means that more rocks that do become stuck cannot pass this point on the wheel
as they are too large to fit through the gap, reducing the chances that the rock will self remove.
Third, the entanglement could be linked with the spacing of the grousers. Currently SherpaTT has
ten double grousers spaced at regular intervals. However as the grousers are doubled this provides
another potential rock entanglement location. These factors are further investigated later in the
thesis.
It was noticed that Curiosity got a rock stuck between its grousers, as seen in Figure 2.2. However,
due to the huge amount of clearance between the wheel surface and any structures above the wheel
the rock was able to continue turning with the wheel until it fell off naturally [18].

Figure 2.2: Rock stuck on one of Curiosity’s six wheels [18]

2.2.2 Rock Entanglement via Wheel Fork

Occasionally when driving during the Morocco field campaign the wheel would become marooned
on larger rocks. This would occur when the wheel fork would become stuck on top of a rock leaving
the wheel spinning freely in the air below. In these situations the only way for the rover to free
itself was to perform a stepping manoeuvre. Whilst it is possible for the rover to free itself from this
type of rock entanglement it is preferable that it is not a situation which occurs frequently.
This kind of entanglement has not been noted during any of the previous Martian rover missions.
However, this is possibly due to the way in which the rovers are designed. SherpaTT has the
ability to turn its wheels through nearly 360 degrees. None of the previous Martian rovers have
this capability. Therefore it was possible to ensure that the wheel fork always came in towards
the rover. As each rover has a certain amount of clearance between its body and the ground it is
important to try and ensure that the rover is not driving over anything that it could beach itself
upon, thus reducing the possibility of a wheel fork entanglement.

15
 Figure 2.3: SherpaTT Wheel Fork Stuck on Rock During Morocco Field Tests

2.3 Past Work on the SherpaTT

The original Sherpa rover featured flexible wheels created by the DLR-RM. Flexible wheels increase
the contact area with the ground, therefore increasing the traction of the wheel and allowing a
narrower wheel to obtain comparable performance to a wider wheel [19]. As the wheel flexes it
matches the ground it is travelling over decreasing the overall stress on the wheel and reducing
the amount of sinkage [20]. Whilst rigid wheels have superior performance on rigid ground the
performance of a rigid wheel significantly degrades on soft ground. The opposite is true for the
flexible wheel as the wheel does not sink as much on soft soils due to the increased ground contact
area [21].
The current SherpaTT also features flexible wheels, however the rigidity is much higher compared
to the previous iteration. The wheels currently used are similar to the proposed flexible wheels by
[20] and flex by approximately 7mm when under the nominal load of 166 kg, as described in more
detail in Section 2.1.3.

2.4 Previous Mars Rovers

As of present four rovers have successfully landed and operated on Mars. These are Sojourner,
Spirit, Opportunity and Curiosity. Each of these have had a number of successes and failures with
their locomotion system which can provide a valuable lesson.

2.4.1 Sojourner

Sojourner was part of NASA’s Pathfinder mission and was the first rover to land on Mars, touching
down on Mars in July 1997 in Ares Vallis. It acted as a proof of concept that it is possible to send
and operate a rover on Mars. Over several months Sojourner travelled approximately 100m, sending
images back via the Pathfinder lander, and opened the path for the future rovers [22].
Sojourner had six wheels of 130mm diameter and 160mm width. Each wheel used stainless steel
spikes to give it better grip on the Martian regolith and the rocker-bogey suspension allows the

16
rover to surmount obstacles larger than the wheel diameter. As the first rover to drive on Mars
even less was known about how the wheels would perform on the Martian surface. To enable the
rover to drive through rocky terrain all the corner wheels could be individually steered allowing
the rover to turn through a small radius of 740mm [23], [24]. The wheels can be seen in Figure
2.4.

Figure 2.4: Figure showing Sojourner rocker and wheel assembly[25]

2.4.2 Spirit and Opportunity

Spirit and Opportunity were two identical rovers sent by NASA as part of the Mars Exploration
Rover (MER) programme as part of a mission to discover more about past water activity on Mars.
Spirit landed in the Gusev crater in January 2004 with Opportunity landing two weeks later on
the Meridiani Planum. Spirits mission lasted for over seven years or 2208 Martian Sols, with the
last communication from the rover received in March 2010 and the final communication sent to
the rover in May 2011. Opportunity’s mission lasted over 14 years or 5352 Martian Sols with the
last communication from the rover in June 2018 and the last communication sent to the rover in
February 2019. Spirit covered a total distance of 7.7km and Opportunity 45.16km, the largest
distance covered by any ground vehicle on another celestial body. Both rovers far exceeded their
initial 90 Sol mission [26], [27].
There were six wheels used on each of the MER rovers. Each wheel was 250mm in diameter
and was milled from a single piece of aluminium. This reduced the amount of excess weight and
stress points where parts would join together as screws or welding points were eliminated. As the
engineers were concerned about rock, dust and debris getting inside the wheels and interacting
negatively with the drive and steering mechanisms the yellow foam, Solimide, was added to fill in
the gaps [28]. Two of the wheels can be seen in Figure 2.5.

Figure 2.5: Figure showing two of the MER wheels with the yellow Solimide infill [28].

Both rovers had relatively few problems with their wheels during their missions. However both
rovers suffered from issues with both slip and sinkage. Opportunity become stuck for 5 weeks

17
in a sandbank ripple that it was attempting to cross. When all six wheels were on the ripple
surface the rearmost wheels were bearing most of the load and began to sink into the sand.
The sinkage, compounded by very high slip, highly deformable terrain and increased compaction
resistance prevented Opportunity from moving forward. After careful manoeuvring Opportunity
was able to free itself and continue on until it was eventually taken out by the largest Martian
dust storm on record. The sand ripple and deep tracks of Opportunity can be seen in Figure
2.6 [29]. Spirit, however, became permanently entrapped in a soft sand area. The area it was
traversing over consisted of highly deformable sulphate rich soils which were covered by solid but
thin layer of basaltic sand. As the solid top layer hid the deformable ground underneath it was not
possible to detect dangerous areas ahead of time. Spirit became temporarily entrapped a few times
before it’s permanent embedding event in the Troy area. After the rover tilted and its front left
wheel became embedded in a sand trap all rescue attempts over the following nine months were
unfortunately unsuccessful. It then continued as a stationary science platform until its mission end
[26], [30].

Figure 2.6: Figure showing the entrance and exit tracks of Opportunity after freeing itself from a
sand dune known as Purgatory Ripple [29]

2.4.3 Curiosity

Curiosity was sent to Mars as part of NASA’s Mars Science Laboratory mission with the aim of
discovering if Mars ever had suitable conditions for life. Landing in the Gale crater in August 2012
Curiosity has currently covered over 21km [31].
At four times larger than its predecessors Curiosity’s wheels were updated from the MER rovers but
still maintain some similarity. The wheels are much larger, at 500 mm diameter and 400 mm width,
in order to carry the increased weight, and have a new grouser design using a chevron pattern.
The chevron pattern provides extra resistance against side slip, thereby improving the wheels
performance when traversing slopes. Additionally the chevron pattern granted a location where
sharp edges could be utilised to aid the rover when overcoming obstacles or climbing up rockier
surfaces. As with the MER rovers the wheel is machined from a singular piece of aluminium which
improves the mechanical properties of the wheel. However, Curiosity’s wheels do not have the
Solomide cover. This allows rock, dust and debris to flow in but also makes it easier to flow out
again [17] [32]. The wheel can be seen in Figure 2.7.

18
 Figure 2.7: One of Curiosity’s six wheels [18]

Curiosity’s wheels are made from hard anodised, aluminium alloy 7075-T7351 [33]. The 7000
aluminium alloy family have zinc as the major added element and often also contain magnesium,
copper and chromium in smaller quantities. These are most often used in aerospace and high
stress applications. The T73 temper provides an excellent stress corrosion cracking and exfoliation
resistance [34]. The grouser design on Curiosity is not based purely on the traction of the wheel. As
the wheels also functioned as the landing gear for the rover the grousers also had to bring structural
elements into the wheel. As such the design was, as it normally is, a compromise between different
areas [17].
A large factor in the Curiosity wheel design was around the weight of the wheels due to launcher
constraints. Consequently the wheels surfaces are only 0.75mm thick, the thinnest it was possible
to machine [17]. This had great weight savings, however it has caused issues with the mission.
During examinations of routine wheel photos that Curiosity sends back, it was noticed that the
wheels were degenerating at rates faster than expected [35] [32], [36]. The first breaks in the wheel
surface were noticed in October 2013, as the rover drove over an area with many sharp rocks
which were embedded in the bedrock [37] and the first breaks in the wheel grousers seen in March
2017 [38]. It has also been noted that tears are forming at the higher stress points at the pointed
tips of the grousers [32]. Whilst it was always expected that the wheels would degenerate over
time due to wear as the rate was alarming work began to investigate why wear was happening so
fast. The cause was found to be a combination of driving over particularly hard and sharp terrain
and from all the wheels driving at the same pace causing wheels to be dragged over obstacles.
After developing a new algorithm and completing successful testing on a model on Earth that has
the equivalent properties of Curiosity on Mars a traction control software has been employed to
increase the wheel life span. By controlling the speed of each individual wheel the chances of any
wheel being partially dragged over an obstacle is reduced. As each wheel can spin faster when
overcoming an obstacle it can keep up with the other unobstructed wheels which have a shorter
distance to travel. In addition driving plans now actively avoid these sharp embedded rock areas
[35], [36].
Curiosity also had a close encounter with a sand trap during Sol’s 709-711 when making an
attempted entry into Hidden Valley. During the descent down the ramp the wheel spin was
increasing sharply as the sand proved to be far more slippery than expected. However the rover
was able to retreat and on Sol 731 it was decided to reroute around the valley [39].

2.5 Previous Lunar Rovers

In addition to the rovers which have been sent to Mars there are also several which have been sent
to the moon. The wheel design for these rovers is quite distinct to those which have been sent to
Mars as the environmental conditions are very different. The moon is covered mostly in a very

19
fine, loosely packed top soil that has a low cohesion. These soil conditions coupled with a gravity
of only 1.62 m{s, compared to the Martian gravity of 3.71 m{s and Earths 9.81 m{s, leads to the
choice of wheels which have a wire mesh structure in order to be as light weight as possible and
to provide an easy exit for lunar soil which ends up inside the wheel. Providing an easy exit for
the lunar soil makes it easier for the wheel to avoid extreme sinkage. Using a wire mesh structure
for the wheels has been possible on the moon as the reduced gravity reduces the stresses that the
wheels will experience. It is desirable to make the wheels as lightweight as possible as it reduces
the payload mass which leads to either a cheaper launch or the allowance to have a larger mass in
other areas of the design. Three agencies have successfully operated rovers on the Moon. The first
was Russia with the Lunokhod rover using a wire mesh supported by thin spokes and a solid rim
and angled grousers to aid traction, as seen in Figure 2.8. Second was NASA with the lunar roving
vehicle as part of the Apollo missions. The wheel can be seen in Figure 2.9 where inner metal
bands are used to support the wire mesh frame of the wheel. Finally the Chinese Yutu rover as
part of the Chang’e mission used a design similar to the Lunokhod wheel but with larger spokes,
as seen in Figure 2.10.

Figure 2.8: Figure showing one of the wire mesh wheels used on the Russian Lunokhod rover [40].

Figure 2.9: Figure showing one of the wire mesh wheels used on the American Apollo lunar rover
[41].

20
 Figure 2.10: Figure showing one of the wire mesh wheels used on the Chinese Yutu rover [42].

2.6 Current Work on Planetary Rovers

As of present there are currently two rovers aiming for a 2020 launch date and one for a 2022.
ESA’s Rosalind Franklin Rover as part of the Exomars mission, NASA’s Perseverance rover as part
of the Mars 2020 mission and China’s HX-1 rover. As there is very little available information on
the HX-1 rover it will not be presented further.

2.6.1 Rosalind Franklin (Exomars)

ESA’s Rosalind Franklin rover is part of the Exomars mission which is a joint venture with Russia
and part of a bigger joint mission with NASA’s Perseverance rover. Rosalind is due to land on Mars
in 2023 [43]. The mission aim is to search for and collect samples that could provide information
about the biological past of Mars. With a targeted mission lifetime of around 7 months, or 218
Martian Sols the rover will be required to drive several kilometres [13].
Rosalind is around the same size as the MER Rovers and will use a similar airbag descent. The
wheel design was heavily influenced by lander imposed size constraints as in order to fit inside the
lander the wheels could not be too wide. The wheels will be 285mm diameter and 120mm wide and
can be seen in Figure 2.11. To overcome the loss of traction from having narrower wheels Rosalind
employs flexible wheels in order to increase the wheel-soil contact area which reduces the ground
pressure [13]. Using flexible wheels also provides other positive effects such as reduced sinkage on
soft ground, as discussed in Section 2.3.

21
 Figure 2.11: Figure showing a detailed view of the Rosalind rovers wheel [13]

Alongside the Perseverance rover Rosalind aims to collect samples of the Martian regolith with the
aim being for a later mission to return them to Earth [43].

2.6.2 Perseverance (Mars 2020)

NASA’s part of the joint venture with ESA and Russia is the Perseverance rover, due to land
on Mars in February 2021 in the Jezero crater. With an expected mission time of at least one
Mars year, or 687 Earth days, the Perseverance rover will study and collect soil and rock samples
to search for signs of micro-bacterial life. The collected samples will then be collected by a later
mission for a return to Earth [44], [45].
The Perseverance rover is heavily based upon the design of Curiosity. However with the information
on the state of Curiosity’s wheels there has been some redesign with the wheels. The wheels will
have a thicker skin to prevent penetration happening so readily whilst also changing the grouser
pattern to a wave to eliminate the pointed edges causing localised areas of higher stress. The
number of grousers has been increased from 24 to 48, with tests showing an improved performance
on sharp rocks without degrading the wheel’s performance on sand [46]. The wheel will also be
narrower but with a larger diameter compared to Curiosity being 25 mm larger with a 525 mm
diameter[47], [48]. The rest of the wheel remains unchanged. The wheel for Perseverance can be
seen in Figure 2.12.

22
 Figure 2.12: Figure showing the wheels on the Perseverance rover [49].

2.7 State of the Art Technologies

In this section various state of the art technologies are examined to provide an overview of where
rover wheels may be headed in the future. Although some of these concepts are already in use
on Earth, or have been used on the Moon, none have been used on any past, present or planned
Martian rovers.

2.7.1 Oragami Wheels

An origami wheel is actively deformable, being able to change its diameter when commanded to
do so. With the wheel fully extended it has the ability to overcome larger obstacles and cross over
larger gaps, when the wheel is folded it is able to fit through smaller gaps. Having the ability to
adapt to its surroundings results in a wheel which is able to cover a variety of rugged terrains that
may be unreachable for a rover with conventional wheels. The difference between an extended and
folded wheel can be seen in Figure 2.13 [50].

Figure 2.13: Figure showing an origami wheel fully extended and fully folded [50]

2.7.2 Wire Mesh Wheels

Wire mesh wheels have been used previously on lunar missions, notably the Lunar Roving Vehicle
from the Apollo missions [41], as discussed in Section 2.5. However, new advances in this area
have produced wire mesh wheels that have the strength and flexibility to withstand Martian and
even Terrestrial usage, utilising the material properties of a titanium alloy and the mechanical

23
properties of the mesh structure. NASA has expressed hopes of using this type of wheel, seen in
Figure 2.14, on some of its future rover missions [40].

Figure 2.14: Figure showing the most recent mesh wheel proposed by NASA [40].

2.7.3 Sphere Rover

The sphere rover wheels work by placing the entirety of the rover inside the wheels, creating a
spherical rover, as seen in Figure 2.15. Each side of the sphere is able to spin independently,
therefore being able to propel the rover forward, backwards and to spin the rover. A more recent
design adds a simple hopping mechanism to the base of the sphere allowing the rover to jump out
of places where it may have become stuck. The simplicity of the design makes this type of rover
is very suited to difficult and rugged terrain. One of the main drawbacks with this rover is its
physical limitations on science. The size of the sphere is limited and the shape of the space inside
does not lend itself well to conventional instruments. The instruments on board would also need
to be robust to ensure that regular hopping or potentially large impact forces from drops do not
damage or interrupt data collection [51].

Figure 2.15: Figure showing a spherical rover [51].

2.7.4 Adjustable Paddle Wheels

Adjustable paddle wheels attempt to combine a walking and rolling motion on the wheels. The
paddle lengths can be changed so that the rover is either entirely resting on the paddles, providing
the walking motion, or so that the paddles are essentially fully retracted and the rover rolls as if on
regular wheels, as seen in Figure 2.16. The advantage of this method is that on hard level ground
the most efficient form of locomotion is a pure rolling motion, however when in soft or sloping
ground longer grousers are preferable. This effective controllable grouser length allows for the

24
wheels to adapt to the situation that they are in and gives the wheels a higher overall performance
[52].

Figure 2.16: Figure showing the various configurations of the paddle wheel [52].

2.7.5 Centre of gravity (CoG) Movement

Using CoG movement takes inspiration from nature where animals will shift their CoG forward
when climbing hills, particularly seen when animals such as camels climb sand dunes. This CoG
shift provides more efficient traction and reduced slip when applied to rovers. The CoG can be
moved in several ways such as using a movable weight to position the CoG in the most favourable
position, leaning a rover with walking capabilities forward as it climbs or by re-positioning any
movable instrumentation [53], [54]. One such method can be seen in Figure 2.17 where a movable
weight is used.

Figure 2.17: Figure showing a potential method of controlling the CoG of a rover [53].

2.7.6 Real Time Sinkage Detection

This concept aims to be able to detect when a wheel is beginning to sink in order to prevent
full entrapment from sinkage. The proposed method uses strain gauges connected to the grousers
to make real time measurements, as seen in Figure 2.18. If the rover detects that the wheel is
sinking it can then adjust itself to prevent further sinkage. By intervening before a wheel has sunk,
scenarios such as the one experience by the Spirit rover, where the wheels became too embedded
in the regolith, can be avoided [55].

25
Figure 2.18: Figure showing how the strain gauges interact with the grousers on the wheel. [55]

2.7.7 Omnidirectional Wheels

Omnidirectional wheels, as seen in Figure 2.19, are already being used in the automation industry
where their ability to move in any direction is a valuable asset. Whilst they have been investigated
for potential use in a rover their inability to carry a high load means they are currently not suitable
for this kind of task [56].

Figure 2.19: Figure showing one possible configuration of an omnidirectional wheel [56].

2.8 Martian Environment

When designing a mission it is important to take into consideration the environment that the
parts will be exposed to. The Martian environment is very harsh with a thin atmosphere, extreme
temperature drops and a terrain that is still not fully understood. Parts used on a rover must
therefore be carefully selected and tested to ensure that they will survive these conditions.

2.8.1 Terrain

The Martian terrain is dominated by dust, sand and rocks in what is broadly known as the Martian
regolith. However, within this broad description are a range of different terrain types. Some areas
are covered in deep sand with rippling sand dunes, others are exposed bed rock, others are rock
debris fields, some are a mixture of all these types. Each type of terrain comes with its own unique
set of challenges. The deformable Martian regolith can be roughly broken down into four different
categories that would likely be traversed by a rover, as shown in Table 2.1 [13]. Each category
has been reproduced on Earth as an Engineering Soil Simulant (ES) to be as close as possible
to conditions on Mars. Rocky terrain such as exposed bedrock is not included in this scale as it
will not deform under the weight of a rover. As no regolith samples have yet been returned from
Mars these reproductions rely on limited data gathered by previous rovers, landers and orbiters
[57].

26
 Table 2.1: Martian Regolith Types and Classification

Regolith Classification Description Occurrence

Type
Fine Dust ES-1 Compressible, very fine grains, Outside of windy areas
porous and in wind shadows.
Fine Sand ES-2 Low cohesion and shear strength Troughs of sand dunes
Coarse ES-3 Moderate cohesion and shear Frequently occurs at the
Sand strength base of bedrock cliffs
Gravel ES-4 Don’t need to worry about Most common terrain
cohesion or shear strength outside of sand dune
anymore. Has a tendency to areas
clump together.

The key difference between the ES types lies within the mechanical properties of the soils. These
being the properties which define the shear strength, τ ; cohesion, C, and angle of internal friction,
φ, and those used in terramechanics; coefficient of cohesion, kC , coefficient of friction, kφ , and the
sinkage exponent, n. If these properties are known it is possible to determine whether the soil
will be able to support any given wheel or rover. As the mechanical properties of the soil change
depending on the humidity and packing density of the soil, amongst other properties, there are
many variations in the potential behaviour within each of the ES types as well [57]. However, until
material can be collected and successfully returned to Earth for testing it still cannot be truly
known how the wheels will behave when traversing the Martian regolith.
Of these four categories it is the lower classifications with the finer sand which are the trickiest
for a rover to traverse. This is particularly true if the soil is only loosely compacted, meaning the
density is low. High slip rates and sinkage are the two most dangerous aspects of driving through
sandy areas as both have the potential to immobilise a rover. These issues often come together as
when a rover is experiencing high slip rates there is a tendency for it to dig itself deeper into the
regolith thus suffering from sinkage too. Each of the Martian rovers, which have driven more than
a few 100m, have faced at least one moment during their mission where the slip rate was seen to
become exceptionally high as the sand is much slipperier than has been predicted [29], [30], [26],
[39]. Excessive sinkage in one dune proved to be the end of the mobile mission for the Spirit rover
[30], [26].
Exposed bed rock eliminates the risk of sinkage entirely however it brings a new risk. Due to the
hardness of the bedrock there is no give in the ground as the rover drives over it increasing the
forces acting upon the wheel. Where the exposed bedrock is jagged and pointy this problem is
exaggerated further as this creates high localised stress on the wheel which has the potential to
puncture the wheel surface [35], [37]. Increasing the stresses on the wheel accelerates the rate of
wheel degradation, reducing the wheel’s lifetime. Boulder fields can also be particularly tricky for
a rover to navigate, as well as posing a large risk to a landing craft. Where there are large rocks
that the rover cannot drive over the rover must go around. This increases the time it takes to drive
to a destination, reducing the number of potential science sites that the rover will be able to reach
[58].
Within these types of terrain regoliths also lie further terrain features. Mars is covered in many
craters of various sizes and ages creating shallow and deep basins with varying amounts of sediments
layered inside. Valleys cut across the landscape, thought to be formed by ancient water and glacial
processes, ranging from shallow valleys that are noticeable from escarpments to the enormous
depths of Valles Marineris at up to 7km deep [59]. Mesa’s and chaotic landscapes are found in
areas thought to have been formed by glacial processes. Finally Mars has many mountains that
are several kilometres high, thought to be formed via either volcanic processes or through sediment
deposits and erosion. This includes Olympus Mons, the largest known mountain in the solar system
at nearly 22 km high [60]. When considering a mission to Mars these features are often of high
scientific interest due to the high likelihood of accessible sedimentary layers and link with a wetter
and potentially more alive past. However these features create challenging terrain and present
steep, often scree covered, slopes.

27
2.8.2 Thermal

The thermal environment on Mars is overall cooler than the Terrestrial thermal environment.
Similarly to the Terrestrial environment the temperature varies depending on the location on Mars.
At the equator temperatures can go as high `38˝ C on a hot day with a low dust opacity, whilst
polar temperatures, at above 60˝ , can go as low as ´130˝ C which is the frost point of CO2 . Daily
temperature differences in a singular location can be as much as 140˝ C. Being exposed to these
low temperatures and large thermal cycles puts a large amount of pressure on the rover systems,
particularly the electronics, and methods must be in place to protect these systems [61].
Being exposed to temperatures this cold can also cause many materials to become brittle and prone
to cracks. Material selection must be done carefully to ensure that parts are still able to function
even at extreme temperatures. If it is not possible to choose a material for a loaded part that
does not become brittle at low temperatures then a heating system must be in place to protect the
part.
Additionally information on the thermal inertia of an area can provide information on the regolith
present. As rocks lose heat slower than fine sand areas that contain a higher percentage of rocks
will have a lower thermal inertia [58].

2.8.3 Weather

Mars has many similarities to Earth when observing the factors which drive Martian seasons and
weather. Like Earth, Mars experiences seasons, depending on how far it is from the Sun and
finds colder temperatures towards the poles and warmer ones towards the equator. Unlike Earth,
Mars has more predictable repeating weather events which can aid mission design. By examining
Martian atmospheric currents it is possible to see that most dust storms occur within the same
seasonal window each year. Storms can happen at any point on the Martian surface, however,
studies have shown that certain areas are more likely to be the starting point. These storms have
the potential to begin a global dust storm event, where dust covers most of the planet for a period
of at least three days. Dust storms block out sunlight which decreases surface temperature and
prevents the ability to gather solar power. Therefore when designing a mission it is important to
take the worst case scenario to ensure that the rover can withstand through any large scale events
that occur within its operating lifetime [62], [63], [64].

2.9 Potential Mission Locations

When deciding upon the wheel design it is also important to consider what kind of locations the
rover might be sent to and therefore what the terrain could be like. Several different locations
were investigated in terms of their likely terrain, scientific potential and landing suitability. The
locations discussed can be found labelled in Figure 2.20 showing where they are on Mars.

28
Figure 2.20: Map of Mars showing potential mission locations (green), previous successful missions
(white), previous failed missions (grey) and planned missions (blue) NASA / JPL / USGS (image);
Emily Lakdawalla (original map excluding the potential mission locations) [65]

2.9.1 Nanedi Valles

Nanedi Valles was first brought in as an area of potential interest with a photograph from the
Mars Global Surveyor in 1998, seen in Figure 2.21. Images showed an inner channel within the
valley which suggests that there was a sustained water flow in this region [66]. Features of Nanedi
include meanders, slip-off and undercut slopes which are all indicative of a surface water flow [67].
Additionally Nanedi is notable for its tight meanders that exist along approximately 50 km of its
length which are generally indicative of perennial flow contradicting evidence of sustained water
flow [66]. Therefore this location would be interesting in terms of the water history of Mars and
for the potentially exposed sedimentary layers in the valley walls and undercut slopes.

Figure 2.21: Photograph captured by the Mars Global Surveyor showing an inner channel within
Nanedi Valles and its location within the system [68]

29
2.9.2 Naktong

Naktong Vallis is part of one of the longest system of lakes and rivers on Mars, stretching over
4700 km, with the Scamander and Mamers valleys being the lower part of the same network. As
it would appear that the source of the Naktong valley is within amphitheatre shaped craters it
implies that groundwater sapping was the main process that was occurring within this system [69].
Similarities have been drawn between Naktong and formations that occur in semi-arid zones on
Earth [66]. Naktong starts at an altitude of 1.8 km at the source and drops down to ´1 km. The
Eastern termination of Naktong Valles can be seen in Figure 2.22. From the ripples present in the
image it can be inferred that there are likely accumulations of softer sand which have been blown
by the wind. These could potentially present issues for a roving vehicle.

Figure 2.22: Photograph captured by HiRISE showing the Eastern termination of Naktong Valles
where the ripples of sand dunes can be seen along the valley floor and inside the nearby crater [70]

2.9.3 Mamers

Mamers Vallis is a highly studied area of Mars where there is evidence of glacial formations and
continued glacial activity. Much of the valley is described as flat and wide, with many mesa’s that
are showing signs that they are still slowly loosing material to the valley floor. This shows that
there are still active processes ongoing [71]. Covering over 1000 km Mamers valley is the end of the
system of lakes and valleys which began at Naktong. Figure 2.23 shows evidence of these active
processes as flow marks can be seen coming from the valley wall and the formations on the valley
floor is thought to be formed by subsurface ice movements and sublimation [71].

Figure 2.23: Photograph captured by HiRISE showing part of Mamers Valles including some of
the valley floor seen on the left side of the photograph and one side of the valley wall seen on the
right side. [72]

30
2.9.4 Marineris

Valles Marineris is the largest valley on Mars at over 4000 km in length, 200 km wide and up to
10 km deep, as seen in Figure 2.24. With a valley floor that is large enough to land a spacecraft
into and exposed sedimentary layers within the valley walls there is a huge amount of scientific
potential. Despite the size of Marineris the mechanisms behind its formation are still unknown. It
is generally believed that the main forces which created the valley were tectonic, however, several
theories exist as to the precise cause [73], [74], [75]. Many parts of the valley have been previously
investigated as potential landing sites for the MER and MSL rovers [76].

Figure 2.24: Image showing the topography of Valles Marineris and the different locations that
reside within it [73].

2.9.5 Nilosyrtis Mensae

Similar to Mamers Vallis, Nilosyrtis Mensae is an area of suspected glacial activity with areas
with exposed sediments, as seen in Figure 2.25 and the potential for ice underneath the regolith
[77]. The terrain is comprised of mesas, cliffs and flat bottomed valleys filled with moraine like
features, glacier like flows and viscous flow features, [78]. Erosion of the valley has left behind
layered deposits, such as those seen in Figure 2.25, which have the potential to provide a history
on the formation of the mensae.

Figure 2.25: Photograph captured by HiRISE showing an area of the Nilosyrtis Mensae where
there are layered deposits [79]

31
2.10 Wheel Terramechanics
Wheel terramechanics is a topic that characterises wheel soil interactions. M. G. Bekker first
proposed models for soil mechanics in 1956 [80], both characterising soil properties and creating
mathematical models describing the rolling motion of a wheel over various surfaces. His work
showed that the performance of a wheel varies depending on the soil properties of the surface it is
travelling over and laid the foundation for the current understanding of wheel terramechanics.
Based upon Bekker’s work further research has been completed which has expanded the model
and addressed several of the limitations of Bekker’s original model. However even these expanded
models still cannot model soil compaction or dynamic movement. Calculations have only been
carried out using the rigid wheel model as there is no analytical model for a flexible wheel on
soft ground that can be used. Though it can be said that a flexible wheel acts like a rigid
wheel when travelling over very soft ground. Therefore whilst these calculations can provide a
reasonable estimation for the wheels performance currently the only way to accurately gauge a
wheel’s performance is through physical testing.
Bekker completed models for four cases a rigid wheel on hard and soft ground and a pneumatic
wheel on hard and soft ground. The first case is most suitable for situations such as a train
wheel running along railway tracks, the second case is suitable for situations such as a cart wheel
travelling over a field, the third case is suitable for a car wheel travelling down a road and the
fourth case is suitable for a car wheel travelling over a field. In the context of a Martian rover it is
the work on a rigid wheel traversing over soft terrain that is of most interest and shall be explored
here.

2.10.1 Rigid Wheel on Soft Ground

When a rigid wheel is on soft ground the ground deforms underneath the wheel causing the wheel
to sink. Depending on the properties of the soil and the design of the wheel the sinkage of the
wheel is variable. This in turn has an effect on the tractive efficiency and draw-bar pull of the
wheel.
The pressure underneath a rigid wheel on soft ground is found by equation 2.1, where k is the
coefficient of proportionality, z is the sinkage and n is a function of the soil properties.

p “ k ¨ zn (2.1)

The coefficient of proportionality can be further broken down as a function of the coefficient of
cohesion, kc , the coefficient of friction, kφ and the width of the surface , b, as shown in Equation
2.2.

kc
k “ pp q ` kφ q (2.2)
b
Equations 2.1 and 2.2 form the basis of the Bekker terramechanics model. In order to calculate the
wheel’s performance prior information must already be known about the soil. This information
is commonly collected using a Bevameter test which presses a block into the soil to simulate a
wheel.
Figure 2.26 depicts a rigid wheel on soft ground. Where W is the load about the wheel, R is the
rolling resistance, D is the diameter of the wheel, b is the width of the wheel, Z0 is the total sinkage
and θ0 is the soil contact angle. As the angle θ increases from θ0 to θ1 it is possible to calculate
the normal force at each angle θ.

32
 Figure 2.26: Forces acting upon a static rigid wheel on soft ground [80]

The draw-bar pull of the wheel, Fd , is the total tractive effort, F , minus the total rolling resistance,
R, as seen in Equation 2.3.

Fd “ F ´ R (2.3)

The rolling resistance can be related to the pressure, in Equation 2.1, by relating it to the
compressive work done, L, in Equation 2.4. This is possible assuming that the ground moves only
vertically. Then, substituting in the rolling resistance, R, from Equation 2.4, for the compressive
work done, L, from Equation 2.5 the resulting Equation which gives R as a function of the sinkage,
2.6, can be found. This can be seen below.

R“L¨b (2.4)

ż z0
z0n`1
L“ p ¨ dz “ k ¨ (2.5)
0 n`1

z0n`1
R“k¨b¨ (2.6)
n`1

Assuming the reaction force, N , acts perpendicular to the wheel, Equation 2.7 can be combined
with knowledge of the wheels geometry, Equation 2.8, to result in Equation 2.9. This denotes
the relationship between the load of the wheel, W , to the wheels diameter, D, width, b, and the
sinkage, z0 .
ż z0
W “´ p ¨ b ¨ dx (2.7)
0

D
AB “ ´ pz0 ´ zq (2.8)
2
?
b¨k¨ D ¨ z0
W “ ¨ z0n ¨ p3 ´ nq (2.9)
3

33
2.10.2 Bekker Limitations

Despite its proven applicability and usefulness in predicting wheel performance the Bekker method
has limitations.
1. The model is for a static sinkage, not for a dynamic sinkage. That is, that it assumes the
soil is only moving in one direction. In reality this is not the case as some soil will also move
laterally. Therefore as soon as the wheel begins to move the model can no longer make an
accurate prediction.
2. The model assumes the maximum force will act upon the dead centre of the wheel regardless
of the sinkage experience.
3. The soil conditions must already be known in order for an accurate prediction to be made.
Given the very sparse knowledge on mechanical properties of the Martian regolith the model
can only provide an estimation.
4. The effect of grousers is not taken into account when estimating the total draw bar pull.
5. The model is ineffective for wheels of small diameter and can provide inaccurate results.
6. The model is based upon a 2D wheel.
Some of these limitations have been addressed by Reece and Wong. The Reece-Wong model
provides a better estimation of the point where the maximum normal force occurs [81]. Another
allows the calculation of grousers to be included [82]. Further work has also been completed using
discrete element methods [83] and more novel methods such as the plastice terramechanics particle
model [84], allowing for a dynamic and 3D wheel to be considered.
In the context of this thesis the simpler Bekker terramechanics model is used as the traction of
the wheel is not the focus of this thesis. Using Bekker’s model allows key parameters such as the
wheel sinkage, rolling resistance and normal and shear forces to be calculated.

2.11 Durability
The durability of the wheels is very important. If the wheels are not durable enough the wheels
will degrade at a rapid rate and become quickly unusable, jeopardising the mission. As previously
mentioned in Section 2.8 the Martian environment is extremely harsh and the wheels must be able
to survive all likely scenarios.
An important balance must be struck between the toughness of the wheels and the weight of the
wheels. Due to launch weight requirements the rover and its wheels need to be as light weight
as possible. However, if too much material is taken off the wheels, they may become too thin
and fragile. Simply using the lightest material is also not an option as the material must be
within the engineering limits of the wheel. Care must also be taken with regards to how brittle
or ductile a material is. Often harder or stronger materials are brittle. A material that is too
brittle will fracture too easily which can lead to the catastrophic failure of a part. A material that
becomes brittle when exposed to extreme temperatures would also be unsuitable due to the very
low temperatures found on Mars. There may also be limits on the cost and manufacturing method
that can be used which also limits the material selection. Therefore a careful study must be made
to select a material that provides the best compromise for all these conditions.
In general, to test the durability of wheels a test bed is used which is filled with material that closely
resembles the expected environment. The wheel is then driven around the test bed repeatedly and
monitored for changes in structure and performance. This often includes a mixture of indoor
testing of singular wheels all the way up to outdoor testing of the whole rover. Singular wheel
testing can be seen as sufficient if it is expected that the rover will only experience quasi-static
loads [36]. However testing using the whole rover or a version of the rover that has the same
mobility system but reduced weight that is the equivalent to what the rover will experience on
Mars is more rigorous.

34
Curiosity wheel testing focused mainly on the landing aspect of the wheels and the lift testing
took place using a singular wheel, this coupled with the lack of knowledge on what terrain to
expect led to unforeseen wear on the wheels. Later testing took place using the Scarecrow model,
seen in Figure 2.27, to aid the engineers in understanding why the wheels were degrading so fast
[36].

Figure 2.27: Figure showing the Scarecrow version of the Curiosity rover which is a stripped down
version of the original to allow Earth based testing of the mobility system [85]

2.12 Summary
By discussing the SherpaTT rover and breaking down the key components and structures that are
investigated in this thesis and looking into what the rover has previously completed. From this the
rock entanglement issues that the rover had during the Utah and Morocco field campaigns were
highlighted and the two situations that can cause rock entanglements were discussed.
Following on from looking into SherpaTT other previous Martian and Lunar rovers were investigated
to see what kind of wheels have been used in the past. This investigation showed that several
different types of designs have been used and each had their own bonuses and drawbacks. The
investigation then moved on to look at rover missions that are currently under preparations
to see what the current state of planetary rover wheels is. From this it was seen that whilst
there is a degree of sticking to what has worked before, for example Curiosity and Perseverance
have very similar wheels, there is also progress in using more novel technology as in the case of
Rosalind Franklin’s flexible wheels. Finally research was completed into state of the art technologies
which are pushing the boundaries of wheel design. This revealed a large assortment of different
technologies, all designed to overcome hurdles that more traditional wheels face. From these state
of the art designs an insight can be made into where rover wheel design may be headed to in the
future as the technology continues to develop.
The next part of this chapter explored the Martian environment, in terms of terrain, temperature
and weather. It showed how even though the Martian environment might seem to be relatively
uninteresting, compared to Earth, there is still a large amount of variation with differing scenery,
soil types, thermal environments and its own unique weather cycle. All of this knowledge is
important to take into account when designing a wheel that must be able to survive these conditions.
Several different potential mission locations were also looked into, discussing briefly their benefits
and drawbacks.
Once the potential types of terrain that could be encountered or locations that could be visited
had been discussed this chapter then examined the terramechanics between the wheel and the soil.
This focused on work by Bekker who’s studies on wheel terramechanics is still used as a basis
for wheel terramechanics models today. The usage and benefits of using the Bekker model were

35
reviewed alongside the limitations and further improvements by Reece and Wong, among others,
were mentioned.
Finally the durability of the wheels was considered and methods of testing the durability were
briefly mentioned. As the rover will be on its own from the moment it leaves Earth the wheels
should be resistant to wear and not be prone to cracking. w

36
Chapter 3

Design

This chapter lays out the different stages of the design process. It begins with the initial ideas
on mission location and potential terrain conditions to provide an understanding of what specific
mission set-ups could be. Next the driving parameters behind the grouser and wheel fork design
are explored before an analytical study to find out what the potential values for the grouser height
and number is completed. Finally, the preliminary design and setup of simulations which aided in
selecting the final design presented in Chapter 5 are explained.

3.1 Mission
In order to redesign the wheels for SherpaTT to prevent rock entanglements it is also important to
consider what kind of environment the rover is expected to work in. Based upon the sites discussed
in Section 2.9 one potential mission site has been selected upon which the design is based. Even
though SherpaTT is capable of traversing complex terrain by utilising its dual walking and driving
motion there are still limitations. For example, whilst a location such as Nanedi Vallis would appear
to be a fascinating location to explore the sides of the valley are noted to be very steep. Even
though SherpaTT is able to navigate more challenging terrain than a more traditional rover there
is still a limit on the maximum traversable slope angle which could make it difficult to impossible
for SherpaTT to enter. Additionally there are several landing site restrictions which make landing
in certain areas impossible.

3.1.1 Assumed Landing Site Restrictions

Elevation

The MER rovers had a restriction of landing at an elevation not greater than ´1.3 km [76], [86], by
the time of MSL this restriction was at ´1 km [87]. Therefore it can be assumed with increasing
technology levels that by the time SherpaTT would launch this restriction could be even less.
This restriction makes landing at a location such as Naktong Vallis difficult, particularly in the
southern end which is at 1.8 km, as mentioned in Section 2.9.2. Even considering a reduced altitude
restriction in the future it still remains a problematic site in terms of elevation for three reasons.
The first is that considering the improvement between the MER rovers and the MSL rover is
300 m with eight years between them it is unlikely that the restriction can be improved by 2.8 km.
Second is that even with the ability to land at higher altitudes it still remains safer to land at
lower altitudes. This is because it allows more time for the spacecraft to slow down as it comes
through Mars’s thin atmosphere. Finally, a simple ballistic landing, as used by the MER rovers,
requires longer to slow down to a suitable landing velocity. Assuming a new landing system is
created by the time SherpaTT is launched it can also be expected that this will be more expensive
than a ballistic landing. Therefore a science target in a higher altitude location such as Naktong
Vallis would have to be significantly better than one at a lower altitude in order to be worth the
increased risk and cost.

37
Latitude

The latitude is restricted to simplify the thermal and power management of the rover. Conditions
in the polar region pose severe restrictions on the amount of solar energy that a rover can
expect to receive. Opportunity and Spirit used 10˝ North and 15˝ South [86] and Curiosity
and Perseverance used ˘30˝ [87], [88]. This restriction would not rule out any of the previously
discussed locations.

Terrain

To ensure a safe landing the terrain should be relatively flat with a low concentration of rocks. An
ideal area is also free from fine grained dust which could hide objects of scientific interest, reduce
rover manoeuvrability and cause unfavourable thermal conditions. Dusty areas can be located by
their uniform high albedo, weak radar echos and low thermal inertia. 135˝ W to 190˝ W is thought
to be covered in dust due to the low thermal inertia of that area. Thermal inertia can also be used
in conjunction with high resolution imaging to find out how rocky an area is. Ideally a landing site
should have a low rock abundance. Low rock abundance means less than 20% or less than 10%
rocks. Traditional rovers are slowed travelling through areas of more than 15% rock coverage [58].
The only site currently at a potentially higher risk of having negative terrain is Naktong Valles
which appears to be on the edge of an area with higher albedo. However, the only way to be
certain would be to obtain and study multiple high resolution images of each potential site.

Landing Ellipse

The landing ellipse is the projected area where the lander will land. As technology is advancing
it is becoming possible to shrink the landing ellipse as a more controlled landing is possible and
predictions become more accurate. An uncontrolled ballistic landing, such as the on used by the
MER rovers, is at least 100km long [76] whereas MSL’s aero-manoeuvring lander is only 25km long
[87]. The landing ellipse should be as free from terrain hazards, such as craters, cliffs, as possible
to increase the chances of a successful landing.

3.1.2 Nilosyrtis Mensae

Nilosyrtis Mensae is located approximately 800 km north of where the planned Perseverance landing
site is, see Figure 3.1. Large parts of this area were investigated when looking for a suitable landing
site for Mars2020.

Figure 3.1: Figure showing a topographical map and other notable locales around Nilosyrtis
Mensae. The square indicates where one particular area of scientific interest lies, Image [89].

38
One potential science target in the Nilosyrtis Mensae region is a layered deposit, as seen in Figure
3.2 c). This area is at an altitude in the region of ´1 km, [90] and would appear to have an area
that could fit in an MSL sized landing ellipse as seen in Figure 3.2 a) and b). The distance between
the landing ellipse and the layered area is estimated to be between 10 km to 35 km depending upon
where inside the landing ellipse the rover ended up. Given that there are also smaller craters and
landmasses on the route to the layers and that the entire area could potentially hold information
on the glacial processes that are thought to have formed this region there is a lot of science that
could be performed on the way to the layers site.

Figure 3.2: Figure showing the location of a potential science target in the Nilosyrtis Mensae region
where each imaged is zoomed in from the previous. The red square represents the area in which
the terrain map is valid, the black circle represents a landing ellipse and the yellow star marks the
potential science location. a) is a zoomed in from Figure 3.1 [91], b) is a terrain map of area inside
red square of a) and c) shows the layers found in the starred area [79].

3.2 Grouser and Wheel Fork Parameters

Before wheel design work could begin several initial concept ideas were investigated to see what
parameters of the current wheel design could be changed in order to improve the performance
when encountering rocks. The parameters identified below were selected after researching what
has worked on previous rovers and what is being designed for use on future rovers. Additionally
lessons have been taken from previous rovers which had issues with their wheels to avoid potentially
running into similar problems. Finally research onto the optimum grouser design has been used to
see what other possibilities there are outside of what has been previously used.

3.2.1 Grouser Parameters

The grousers have several parameters which can be changed. These are the angle of the grouser
normal to the wheel surface, the spacing in between each grouser, the height of each grouser, which
can all be seen in Figure 3.3, the pattern of the grousers and the shape of the grousers leading
edge, which can both be seen in Figure 3.4.

39
 Figure 3.3: Figure showing how the angle, spacing and height of the grousers can be altered.

Figure 3.4: Figure showing how the shape and pattern of the grousers can be altered.

Changing the grousers can have a reasonably large impact on the overall performance of the wheel
so it is important to consider what these effects will be when looking at reducing the chances
of rock entanglement. The overall performance of the wheel should not be decreased by these
changes therefore studies will be conducted into the grouser choices with the aim of maintaining
or improving the current traction.

40
Angle

The current SherpaTT wheels have grousers at a 45˝ angle normal to the wheel, this is different
to all the past, current and planned Martian rovers which all use grousers at 90˝ to the wheel, see
figures in Section 2.4 and 2.6. It is thought that changing this angle to 90˝ should prevent rocks
getting stuck under the grousers which would eliminate one of the entanglement processes entirely.
However if the current grousers were changed to still be a double grouser with two 90˝ bends
then this would open the possibility for rocks to become stuck in the centre of the double grouser.
Therefore if the angle of the grouser is changed then the grousers must also be split into single
grousers to ensure that there is sufficient space between the grousers to prevent rocks becoming
stuck.

Spacing

When the grousers are evenly distributed the spacing of the grousers is dependent on the number
of grousers and the size of the grousers. Currently the grousers on SherpaTT are 36˝ apart from
each other. Increasing the number of grousers would therefore also reduce the spacing between
each grouser. Depending on the size of this spacing and the height of the grousers this could lead
to either increased or reduced chances of rock entanglement. Grousers that are spaced very close
together would prevent rocks over a certain size from getting stuck as they simply would not fit.
Equally grousers spaced very far apart would be unlikely to have the grip to lift a rock large enough
to become stuck between them.
Another concern when altering the number of grousers is the wheel’s traction. Grousers provide
extra traction for a wheel by creating additional shear forces within the soil that act to propel
the wheel forwards or act as extra grip on rocky surfaces. Reducing the number of grousers can
therefore have a detrimental effect on the wheels traction. Increasing the number of grousers has
been found to act with diminishing returns, meaning that at a certain point the amount of extra
traction that can be gained from adding a grouser is not worth the additional costs in weight
[14], [15], [16], [92]. Additionally if the number of grousers is increased by an extreme amount the
spacing in between each grouser becomes almost negligible.
The spacing of the grousers is linked closely with the height of the grousers in terms of wheel
performance. In general if a wheel has many short grousers and the grouser number needed to be
reduced without affecting the tractive performance of the wheel this could be achieved by increasing
the height of the grousers. In this way a wheel with many short grousers can have the same tractive
performance as a wheel with a few tall grousers [15], [16].

Height

As previously mentioned the effect that the height of the grousers has on the tractive performance
of the wheel is linked with the grouser number. However grouser height should also take into
account the type of soils the wheel is likely to be driving over as changing the height of the grouser
changes the way the wheel interacts with the surface. Extremely tall grousers act more like paddles
and are generally best suited for soft, sandy terrain where soil cohesion is low. Extremely short
grousers start to approximate having no grousers at all and work best on hard flat ground with a
high soil cohesion. When grousers are fully filled with soil the grousers then act as if the radius of
the wheel has been increased by the height of the grousers. Testing by Gao et al [14] and Sutoh
et al [92] has shown that changing the grouser height can have a larger difference in terms of
drawbar pull than changing the number of grousers, however there is a practical limit to how tall
the grousers can be made.
In general most wheels use a grouser height between 2-25mm [15]. The current effective grouser
height on SherpaTT is 9mm when the perpendicular distance between the grouser base and grouser
tip is measured.

41
Pattern

The pattern used by the grousers on the wheel surface affects the wheel’s resistance to side slip
by increasing the lateral forces [93], particularly when driving alongside incline. The use of side
grousers on the edge of the wheel can also act to increase side slip resistance through the creation of
lateral forces [94]. Grouser patterns have also been used in the past to act as an odometer, where
the pattern is deliberately uneven making it possible to count the number of wheel revolutions
made across a distance [95]. However adding in a grouser pattern can have negative effects as
well. For example on Curiosity having a sharp corner in the middle of the wheel surface acted as
a stress point, causing small failures in the wheel around these areas [32]. Creating more complex
grouser patterns also increases the complexity of manufacturing thus increasing cost. If the wheel
is machined from a single piece of metal this increase in complexity does not impact the wheel in
terms of weight, however if the grousers are to be independently attached to the wheel an increase
in weight is likely to occur due to the increase in complexity of the grouser attachment method.
Finally having grousers which are inclined can produce negative sideways forces and additional
steering torque [14].

Shape

The shape of the grouser does not play a large role when traversing over soft, low cohesion ground as
the other features such as the grouser height and spacing will dominate the performance. However,
if the wheel is also traversing hard or rocky ground then the grouser shape can aid the wheel’s
performance. Adding a sawtooth edge to the grouser makes the grouser act in a similar way
to a crampon or running spike, allowing the grouser to bite into the surface a little to provide
a more secure grip by biting into the harder surfaces [17]. This reduces slip, particularly when
climbing slopes. However, similar to crampons or running spikes, over time the sawtooth edge will
be worn down until the surface is eventually blunt. Therefore depending on the mission terrain,
mission time and grouser material adding a sawtooth edge may or may not be worth it. Adding
the sawtooth adds an extra level of manufacturing and potentially a small amount of extra take
off weight. The current SherpaTT grousers use sawtooth edges which are worn down and need
replacing approximately once every 10km, assuming similar terrain to that found in the Utah and
Morocco field trials.

Overall View

Therefore, the ideal grouser set up is dependent on several competing factors including; soil type,
slope sizes, wheel size, rover weight, constraints on wheel weight and constraints on wheel size
amongst others. Depending on the scenario the rover is expected to travel in different set-ups may
be required.

3.2.2 Wheel Fork Parameters

To prevent rocks from becoming stuck under the wheel fork several potential changes could be
made. These are changing the angle of the wheel fork, (b) in Figure 3.5, the clearance between
the wheel and the wheel fork, (c) in Figure ??, the material, the thickness, (f) in Figure 3.5, and
potentially rerouting some electronics such as the torque sensor and the wires for the wheel motor,
(d) and (e) in Figure 3.5.

42
Figure 3.5: Figure showing how the angle (b), clearance, thickness (f) and electronics (d/e) could
be altered on the wheel fork where (a) shows the original configuration.

Clearance

By increasing the height of the wheel fork it is possible to increase the distance between the top of
the grousers and the wheel fork, therefore providing a larger clearance. In this way even if a rock
was to become entangled within the wheel’s grousers it would have sufficient space to pass between
the gap separating the wheel and the wheel fork until the rock falls off naturally. As shown in
Figure 2.2 a small rock stuck in between the grousers should not cause issues with driving and will
fall off naturally if it can fit through the wheel clearance.

Angle

Changing the angle of the wheel fork would put a greater clearance between rocks and the wheels.
If a rock were to come into contact with the wheel fork it would encourage the leg to slide off
the rock. There is also less of the wheel fork that is low down enough to get stuck in the first
place. This could however have the effect of increasing the overall width of the wheel fork, which
could cause problems transporting and launching the rover if it exceeds the maximum overall
dimensions.

Material

Changing the material of the wheel fork to make it either have a lower friction value or slightly
more give could reduce the chances that it gets stuck on a rock. Changing the material, however,
could have negative consequences on the rover as the wheels must still support the full weight of

43
the rover. The current SherpaTT uses a metal alloy that is moldable and also corrosion resistant.
However the new design will not be limited to only corrosion resistant materials.

Thickness

Reducing the thickness of the wheel fork would reduce the chances of rock entanglement by
probability. With less surface area that could be mounted on a rock it is simply less likely to
happen. However, in order to do this the torque sensor would have to be relocated to above the
wheel. This would come with the drawback of making it harder to know if the wheel is stuck on a
rock in the first place. Additionally it would have to be verified that the thinner wheel forks could
still support the rover and all the stresses and strains of driving.

Re-routing electronics

Currently there is an extra hub on the outside of the wheel fork which protects two wires that link
the torque sensor and wheel drive motor to the rest of the leg. If these wires could be rerouted,
possibly even inside the wheel fork itself this could also help aid in reducing the thickness of the
wheel fork.

3.2.3 Durability

Durability will be mostly considered in terms of the wheel surface and the grousers. Factors to
be kept in mind when designing parts are the materials used, the thickness of the wheel skin, the
grouser usage and the stresses placed on the wheel. The material must not degrade at a rapid
rate and must be able to withstand the thermal stresses experienced during the whole mission
lifetime. The thickness of the wheel skin is a careful balancing act between the weight of the wheel
and resistance to punctures and tears. The amount of grousers used can also affect the durability
of the wheel as more or less of the wheel skin can be exposed to the ground. The durability of
the grousers themselves must also be high enough to prevent loss of traction due to grouser wear.
Finally, reducing the stresses on the wheel by ensuring grousers are used appropriately and that
any changes to the wheel fork do not increase the stress on the wheel or leg beyond what can be
sustained.

3.3 Grouser Equation

This thesis uses the Grouser Equation as developed by Moreland in [15] and [16] as a basis for
the grouser design. The grouser equation calculates the minimum number of grousers required to
ensure that the outer rim of the wheel does not touch the ground before one of the grousers sweeps
away the soil in front. Sweeping away this soil before the outer rim touches it reduces the resistive
forces acting on the wheel.
In order to complete the grouser equation some basic information about the wheel’s interaction with
the soil must be known. By utilising the Bekker theory of terramechanics, introduced in Section
2.10, and by assuming that the soil properties will be the same as the ESA Martian regolith types
[57], as found in Table 3.1, it is possible to obtain this information.

Table 3.1: ESA Martian soil properties [57].

Soil Type n kc pk{P a{mn q kφ pk{P a{mn´1 q

Fine Dust, Low Density (ES-1) 0.67 67.28 0.68
Fine Dust, Medium Density (ES-1) 0.71 61.96 1.30
Fine Dust, High Density (ES-1) 0.75 142.36 1.66
Coarse Sand, Low Density (ES-3) 0.92 1727.51 -14.12
Coarse Sand, Medium Density (ES-3) 0.87 1931.13 -16.41
Coarse Sand, High Density (ES-3) 0.76 2312.59 -30.10

44
3.3.1 Bekker as a basis

Starting with the Bekker relationship between the weight acting on the wheel and the sinkage, as
shown in Equation 2.9, it is possible to rearrange the equation to solve for the sinkage. This can
be seen in Equation 3.1
d
pn´ 1 q
3¨W
z0 “ 2 ? (3.1)
b¨k¨ D ¨ p3 ´ nq

By then taking the soil coefficient of cohesion, kc , soil coefficient of friction, kφ , also known as the
friction angle, and n, which is the sinkage index, and combing these with the known parameters of
the rovers weight, W , wheel diameter, D, and wheel width, b the equation can be solved.

Figure 3.6: Graph showing how the rolling resistance varies with sinkage where the sinkage is found
from Equation 3.1 and the rolling resistance from Equation 2.6. The points are (a) coarse sand,
high density, (b) coarse sand, medium density, (c) coarse sand, low density, (d) fine dust, high
density, (e) fine dust, medium density, (f) fine dust, low density.

Figure 3.6 shows how the rolling resistance increases with the sinkage in an approximately linear
way. The sinkage for the coarse high density sand is the lowest whilst the highest sinkage comes
from the fine medium density sand. This discrepency between the low and medium density fine
sands comes from the way in which the soils were set-up, proving the difficulty in predicting accurate
soil parameters for unknown soils such as those found on Mars. The values are still included as
they are still valid parameters.

3.3.2 Grouser equation from Bekker

Once the static sinkage has been calculated it is then possible to find the optimum grouser
arrangement on the wheel. In addition to the sinkage the wheel radius, r, grouser height, h,
wheel sinkage, z, and wheel slip, i must be also known. The slip is taken to be 20% for this case,
as found during the rover field trials. All values, excluding the slip, must then be normalised in
relation to the wheel radius so that the wheel radius becomes 1, the grouser height ĥ and the wheel
sinkage ẑ.
The grouser equation can be found in Equation 3.2.
b
1 a
φă ¨ p p1 ` ĥq2 ´ p1 ´ ẑq2 ´ 1 ´ p1 ´ ẑq2 q (3.2)
p1 ´ iq

45
The minimum number of grousers, gn , can then be calculated as seen in Equation 3.3

2¨π
gn “ (3.3)
φ

Of the six parameters required to solve the grouser equation three have been left open to wheel
and mission design choice. The first of these is the soil type. As the soil type parameters change so
does the sinkage, which in turn affects the minimum number of grousers. Second is the height of
the grousers. Third is the number of grousers. The parameters which remain fixed are the wheel
width, wheel diameter and the weight acting on the wheel. Fixing the wheel width and wheel
diameter ensures that SherpaTT will still have the same sized minimum footprint so that it can
fit into it’s existing travel containers. The equation was therefore iterated over six different ESA
Martian soil types, detailed in Table 3.1, and completed for grouser heights of 5 mm up to 30 mm.
The results of this plot can be seen in Figure 3.7, with points comparative to SherpaTT’s current
configuration on the plot found in Table 3.2.

Figure 3.7: Graph showing how the grouser height varies with the grouser number and regolith
type.

Overall it can be seen that the relationship between the grouser height and the grouser number
is an exponential one. As the grouser height decreases the number of grousers must increase, but
with diminishing returns. As more and more grousers are added the amount which the height of
the grouser decreases becomes less and less. Therefore, it is not a case of simply adding more
grousers and a more optimum point can be found based upon what works best with the wheel
design. In terms of the relationship with the soil type it can be seen that as the soil becomes finer
grained and less dense either the number of grousers or the height of the grousers must increase.
This is due to the increased amount of sinkage that is found when using these soil types.

46
Table 3.2: Grouser equation results showing what the predicted optimum height and number of
grouser would be if the SherpaTT current grouser height or grouser number was kept the same.
The predicted values are shown in italics.

Soil Type Grouser Height (mm) Grouser Number

Fine Dust, Low Density 9 40
Fine Dust, Low Density 20.5 20
Fine Dust, Medium Density 9 44
Fine Dust, Medium Density 22 20
Fine Dust, High Density 9 37
Fine Dust, High Density 19 20
Coarse Sand, Low Density 9 27
Coarse Sand, Low Density 13.5 20
Coarse Sand, Medium Density 9 25
Coarse Sand, Medium Density 12 20
Coarse Sand, High Density 9 22
Coarse Sand, High Density 10 20

The comparative data points shown in Table 3.2 show what the grouser equation calculates the
minimum grouser number to be if the SherpaTT’s grouser height of 9 mm was kept and what
the grouser height would be if the SherpaTT’s grouser number of 20 was kept, for each soil type.
Comparing the data it can be seen that for the coarse high density sand there is little difference
between what is calculated and what is currently installed on SherpaTT. However as the sinkage
increases the discrepancy between the two also increases.
Based upon these results it can be seen that the optimum configuration depends on several factors
including the soil type in which the rover will spend the majority of its time, the maximum total
size, inclusive of grousers, that the wheel may be and the way in which grousers will be attached to
the wheel which can effect the maximum number of grousers that are possible. There is also a large
limitation in this method when comparing the results with the SherpaTT’s current wheels as this
method gives a guideline for rigid wheels whereas SherpaTT has flexible wheels. However, given
the lack of an analytical method for flexible wheels this output shall still be taken as a guideline.
A study by Iizuka et al [96] has shown that the addition of grousers on flexible wheels still brings
the same benefits as it does to a rigid wheel in terms of traction and slip resistance, however, as
a flexible wheel will not sink as much as a rigid wheel would in the softer soils it can be inferred
that at least for the softer soil conditions the minimum grouser number would be less than the
minimum grouser number required for a rigid wheel.
Table 3.3 shows the suggested values for the grouser number and grouser height for four different
setups. As a rigid wheel is overall more suited for a mission which is traversing terrain that is
generally harder it shall be considered for the ES-3 and ES-2 case whilst the flexible wheel which is
more suited for traversing softer terrain shall be considered for the ES-2 and ES-1 case. It should
also be remembered that there is an ES-4 case which is harder than the ES-3 case, see Section
2.8.1. Even though it may appear that the suggested grouser configuration for the flexible wheel
on terrain type ES-2 is exactly the same as the current SherpaTT configuration it should be noted
that this result is for evenly space single grousers. As SherpaTT’s double grousers have the same
overall spacing as ten single grousers they are further apart than is recommended by this thesis.
From examining Figure 3.7 it can be seen that if the total number of grousers is ten then the height
of the grousers must increase dramatically.

Table 3.3: Table showing suggested grouser parameters for four cases.

Wheel Type Terrain Type Grouser Number Grouser Height

Rigid ES-3 28 9mm
Rigid ES-2 20 15mm
Flexible ES-2 20 9mm
Flexible ES-1 20 15mm

47
3.4 Preliminary Designs
Preliminary sketches were to investigate various different design ideas. The grouser design began
with an investigation into the various different patterns and shapes that were possible. During
this process it became clear that designing a way in which to attach the grousers to the wheel
was going to put a limit on the complexity of the grouser shape. Several methods were considered
including a wrap around attachment, seen in Figure 3.8, an inner support grouser, seen in Figure
3.9 as well as considering welding the grousers onto the wheel or simply machining the grousers
and the wheel surface from the same piece of metal.

Figure 3.8: Grouser that uses a fold at each end to wrap around the edge of the wheel surface
where it is secured in place with a screw.

Figure 3.9: Grouser that uses a second piece of metal on the underside of the wheel surface that
pins it in place.

More complex patterns such as a chevron, seen in Figure 3.10, a grouser with a zig-zag shape,
as seen in Figure 3.11, or a wave grouser, as seen in Figure 3.12, would require a more complex
attachment and would also be more difficult to manufacture than a simple straight grouser, such
as the one in Figure 3.9. The degree to which these patterns are used also affects performance.
For example if a straight grouser is said to be at 0˝ then, in the case of a chevron grouser the angle
can be adjusted up to a maximum of 90˝ at which point you have grousers that run parallel with
the wheels. As the traction benefit of having the grousers decreases as the grouser angle increases
most angled grouser designs will not exceed around 30˝ .

Figure 3.10: Chevron shaped grouser attached via one long tab running along its length.

The zig-zag pattern is similar to the chevron pattern but is smaller in width as the chevron repeats
itself along the length of the grouser.

48
Figure 3.11: Grouser that has a zig-zag shape and is attached using metal tabs on either side of
the grouser.

When looking a grouser with a wave pattern a similar approach can be seen with increasing the
amount of curve. As a wave grouser is essentially a ziz-zag grouser but without sharp edges the
benefits and downsides of increasing the peaks is similar. The wave grouser has the advantage of
not creating localised areas of high stress that the sharp edges of the chevron and zig-zag grouser
do.

Figure 3.12: Grouser that has a wave pattern and small tabs to attach it to the wheel surface.

It was decided to use either a straight or simple chevron shaped grouser for several reasons. The first
is that while grousers with patterns provide increased side slip resistance they have a decreased
traction compared to straight grousers as not all of the grouser is acting perpendicular to the
direction of movement. The second is that on the flexible wheel adding a grouser with any kind of
pattern would inhibit the wheels ability to deform as it increases the size of the area on the wheels
surface which is held in a rigid position, this can be seen by comparing a) and b) in Figure 3.13.
Whilst an example of a flexible chevron grouser can be found, in Farhat et al (2011) [96], it utilises
a much more complex structure to achieve this. Using a straight grouser still affects a wheels
ability to flex as it is still adding another rigid part onto the wheels surface but this is reduced
compared to the chevron grouser which locks up a larger area. Finally it enables a simpler method
of manufacturing which provides savings on cost and increases the ease of maintenance. This final
reason is also the reason it was decided to use screws to attach the grousers over machining them
into the wheel surface or welding them. By using a screw attachment it is possible for single broken
or damaged grousers to be replaced without having to replace the entire wheel.

49
 Figure 3.13: The initial wheel designs using a chevron pattern a) and a straight pattern b).

3.5 Durability Investigation

The current grousers on SherpaTT are made from aluminium 6068 (Al 6068) and utilise a sawtooth
edge. Al 6068 has very similar properties to the more commonly found Al 6063 with the main
difference between the two being the higher magnesium content found in Al 6063. Al 6068 can be
extruded, forged and rolled but not cast during manufacturing. Whilst it cannot be work hardened
it is possible to heat treat it to increase the strength at the cost of ductility [97].
During field testing of SherpaTT it has been observed that after driving for approximatley 10km
the sawtooth has worn away and the grousers require replacing. The speed of the wear is indicative
of the hardness of the grousers. A harder material will wear away at a slower rate compared to
a softer one. However, with increasing hardness there often comes a decrease in ductility and
manufacturability.

3.5.1 Material Properties

The investigation into different materials and their properties can be seen in Table 3.4. Three
main metals are investigated, aluminium, steel and titanium. Steel is the heaviest of the three at
nearly three times the density of aluminium and twice that of titanium. Additionally steel suffers
from corrosion and from brittleness under cold temperatures. Titanium is approximately twice
as dense as aluminium however it is significantly stronger meaning less titanium is required to
achieve the same strength as aluminium. However with an increase in strength comes an increase
in manufacture costs as titanium is harder to work with. This manufacturing cost coupled with
the relative rarity of titanium compared to aluminium also makes it significantly more expensive.
Titanium has excellent corrosion resistance Aluminium is an extremely lightweight, easy to work
with metal that appears in abundance on Earth. It has good corrosion resistance and does not
degrade under cold temperatures, showing an increase in tensile strength.

50
 Table 3.4: Mechanical Properties of Several Considered Grouser Materials [34], [98], [99]

Youngs Fatigue Shear

Density Hardness UTS
Material Modulus Strength Modulus
(g/cc) (vickers) (MPa)
(GPa) (MPa) (GPa)
Al 2014-T6 2.8 155 483 73.1 124 28
Al 2024-T6 2.78 142 427 72.4 124 27
Al 2219-T62 2.84 130 414 73.1 103 27
Al 2124-T851 2.78 146 483 73.1 125 27
Al 5052-H38 2.68 87 290 70.3 138 25.9
Al 6063-T6 2.7 83 241 68.9 68.9 25.8
Al 6061-T6 2.7 107 310 68.9 96.5 26
Al 6082-O 2.7 35 130 70 - -
Al 6082-T4 2.7 75 260 70 - -
Al 6082-T6 2.7 100 340 70 - -
Al 7075-T6 2.81 175 572 71.7 159 26.9
Al 7075-T7351 2.81 155 505 72 150 26.9
AISI 304 8 129 505 193 - 86
6A1-4V-Ti 4.43 349 950 113.8 240 or 510 44

However, as mentioned above, the material’s behaviour when exposed to low temperatures, down
as low as ´130˝ C [61] must not be brittle. Whether a material will become brittle depends upon
its internal chemical structure, its internal chemical composition and the manufacturing processes
used. If the internal structure is face-centred cubic crystals, such as nickel, copper and aluminium,
the metal is likely to remain ductile. If it is body-centred cubic crystal structures, such as iron,
chromium and molybdenum, then the metal is likely to have reduced ductility. Finally if it is a
hexagonal crystal structures, such as magnesium, zinc and titanium, then the metal is likely to
be brittle even at warmer temperatures, with the exception of titanium. However titanium can
become brittle if impurities such as oxygen, or carbon exist inside. Considering these internal
structures is particularly important when using alloys as even though steel itself becomes brittle,
by alloying the steel with nickle the temperature at which brittle fracture can occur is reduced
[100], [101]. Metals that are noted to be suitable for temperatures down to the lowest temperatures
expected on Mars can be found in Table 3.5.

Table 3.5: Mechanical Properties of Metal Alloys which Remain Ductile at Cryogenic Temperatures
[34], [98], [99], [102]

Youngs Fatigue Shear

Density Hardness UTS
Name Modulus Strength Modulus
(g/cc) (Vickers) (MPa)
(GPa) (MPa) (GPa)
Al 2014-T6 2.8 155 483 73.1 124 28
Al 2024-T6 2.78 142 427 (Min) 72.4 124 27
Al 2219-T87 2.84 149 476 73.1 103 27
Al 5456-H343 2.66 101 435 71.0 - 26
Al 6061-T6 2.7 107 310 68.9 96.5 26
Al 7039-T6 2.74 140 400 69.6 160 26
Al 7475-T7651 2.81 162 531 71.7 - 27
ASTM 310 7.89 222 550 196 - -
A-286 7.92 - 620 - - -
5Al-2.5Sn-Ti 4.48 349 861 110-125 290-530 48
6A1-4V-Ti 4.43 349 950 113.8 240-210 44

3.5.2 Potential Materials

Based upon the research detailed above and the studies on previous Martian rovers, there are
several different alloys that are of a higher interest. Consideration is also taken into how easy each

51
of the various metals is to work with.
Al 2014-T6: Has the same hardness rating as Al 7075-T7351 however aluminium alloys from the
2000 series are easier to machine than those from the 7000 series. Corrosion resistance is
poor [103].
Al 6061-T6: Has a lower strength and hardness rating than the other alloys listed here however
it is an extremely easy alloy to work with. It can be easily extruded, formed, welded and
machined and also has good corrosion resistance [104].
Al 7075-T7351: This is the alloy used on Curiosity. It has a good strength to weight ratio and
a high hardness level. Additionally it has good corrosion resistance [105], [33].
Al 7475-T7651: This has the highest hardness of all the aluminium alloys investigated. With
a very good machinability but low weldability and formablity it is a reasonable choice
for manufacturing. It has good corrosion resistance and responds well to being placed in
extremely cold temperatures [106].
5Al-2.5Sn-Ti: Has an excellent hardness and strength rating compared to the aluminium alloys.
Difficult to machine but can be easily welded, formed and annealed. Also has excellent
corrosion resistance [107].

3.5.3 Additional Hardening Methods

Additionally there are several different methods of hardening the material used for the grousers
which will also increase the durability of the edges to wear.
The first is by anodising. Anodising increases the depth of the protective oxide layer on the
surface of the material. By adding this protective oxide layer the length of time that the material
is protected is prolonged as it has increased protection against both corrosion and general wear. A
material is anodised by using the material itself as an anode and passing a direct current through an
electrolytic solution. As the current passes through the solution it releases oxygen at the surface of
the anode which then reacts with the material serving as the anode and builds up an oxide layer up
to 250 µm thick. Hydrogen is released at the cathode. Thicker layers of anodised material are often
porous so in which case they require sealing in order to be corrosion resistant. Anodising is most
commonly done using aluminium but can also be done using titanium [108], [109], [110].
The second is by annealing. Annealing takes place during the metal processing and uses heat to
change the metal’s physical properties. By taking advantage of the fact that the material properties
of most metals are largely dependent on the grain size and phase composition the metal is heated
beyond its crystallisation point and held there at a set temperature and time before the metal is
cooled at a rate suitable for achieving the desired material properties as it recrystallises. Cooling
the material quickly will increase hardness and cooling the material slowly will increase ductility.
Annealing also helps to eliminate defects which may have been introduced to the metal if it has
been deformed [111], [112].

3.6 Summary
This chapter began by narrowing down the potential mission locations by considering the restraints
that are imposed by the landing system. Based upon these constraints and upon the potential for
science the locations were narrowed down to one location, Nilosyrtis Mensae. This location was
then investigated in greater detail with the terrain further explored.
After that this chapter discussed what the different parameters were, in terms of the grouser and
wheel fork design, that could be changed. The benefits and drawbacks of changing each parameter
were discussed and potential solutions began to emerge. Once the parameters were identified an
analytical study was then carried out using a combination of Bekker’s theory of terramechanics
and Moreland’s grouser equation to determine what the optimum number of grousers would be for
the wheel. Several different soil types were considered in this study as the results for a softer soil
are quite distinct from those for harder soils.

52
With the results from the parameter discussion and the analytical study several preliminary designs
were drawn up. These designs were evaluated in terms of how suitable they were for the wheel,
how the different grousers could be attached to the wheel and how difficult they would be to
manufacture. Based upon these evaluations decisions were made on which would be used in the
final designs.
Finally an investigation was made into the material properties of several different aluminium,
titanium and steel alloys in terms of their durability. Considerations included the strength,
hardness and workability of each of the materials. Research was also made into the behaviour
of these alloys at the cryogenic temperatures that can be expected on the Martian surface. As
many metals become brittle at these temperatures it is important to select a material that does
not become brittle at very low temperatures. This is because a brittle failure is far more likely to
end in a catastrophic wheel failure that a ductile or plastic failure.

53
Chapter 4

Simulation Results

As the design of the wheels progressed, new parts were regularly tested using Solidworks simulation
software to find any weaknesses in the design early and to see where there is a potential to improve
upon the design. Simulations are carried out using the Solidworks static case. Both single parts
and the full rigid wheel assembly are tested. The full flexible wheel is not tested as major changes
are not made to the structure of the wheel. In all figures green arrows correspond to points on the
model that are fixed and purple arrows correspond to where a force is applied to the model.
Each new part that was created is tested individually to ensure it can withstand the stresses which
it will be put under. The individual parts tested are the wheel spokes for the rigid and flexible
wheel, the inner wheel rim, and the grousers. Other parts that are not tested individually are
either non-critical, unchanged from previous designs or not possible to test without the interaction
from other parts. To form the mesh a curvature based model is used as it provides a better meshing
quality on curved surfaces compared to the standard mesh. Additionally the h-adaptive model is
applied where possible to ensure that the mesh is at a suitable density. Running the simulation
with the h-adaptive model applied means that the simulation runs multiple times, with the mesh
density increasing at areas of the model which are not yet converging on each iteration, until the
solution converges to at least 97% accuracy. Higher accuracy is not used due to a significant
increase in computation time and limitations in the available computer memory.
The force used in each simulation is 850 N. This is used taking into account that SherpaTT’s
nominal weight is 166 kg and should have a payload capacity of at least 80 kg [7]. Considering
that SherpaTT also has the ability to run on only three legs the total maximum weight each wheel
should experience is 82 kg. Using Martian gravity, 3.711 m{s2 , this results in a force of 310 N per
wheel. However, the rover must also be able to withstand the forces of Earth’s gravity, 9.81 m{s2 ,
which results in a maximum force of 804 N. To add in a further margin of safety, in case the rover
is on a slope which would increase the forces on the lower wheels, the final force used is 850 N. For
the wheel fork studies a torque of 60 Nm is applied to model the steering torque [7].
The results of the simulations show where the maximum von Mises stresses are occurring on the
different parts of the wheel, as well as giving an insight into the current safety factor of the
design. The maximum von Mises stress should never go above the yield stress of the material
as this indicates that the material is failing. Ideally the safety factor should be at least a factor
of 2 so that there is space for maximum loads that are higher than expected without the wheel
failing.

4.1 Flat Grouser

The flat grouser consists of a single piece of metal that has been folded by 90˝ . It is used on the
flexible wheel for the soft soil as in this case the wheel is unlikely to be travelling over many hard,
rocky surfaces reducing the benefits of using a sawtooth edge. This grouser is also the simplest to
manufacture.

54
4.1.1 Simulation Set-up

The grouser has its lower surface fixed, as if it was touching the wheel surface, and is loaded with
a 850 N force in a downwards direction on the top surface of the grouser edge. This can be seen
in Figure 4.1. During the testing on the grousers it was noticed that a singularity was forming
along the steel sheet part bend when the model was treated as a solid part. Therefore the model
is treated as a sheet metal part, this comes with the limitation that adaptive modelling cannot be
used. Instead the mesh was manually refined until the results were converging to 97% to match
the h-adaptive method.

Figure 4.1: Image showing the set-up of the flat grouser prior to simulation.

4.1.2 Simulation Results

The maximum stress found was 1.406 ˆ 107 N {m2 which occurs along the edge of the bend that
connects to the base of the grouser, as seen in Figure 4.2. This is below the yield strength of either
of the recommended materials with the Al 7075 having a yield strength of 3.590 ˆ 108 N {m2 and Al
6061-T6 is 2.750 ˆ 108 N {m2 . The factor of safety is therefore 26 for Al 7075 or 20 for Al6061-T6.
Where the safety factor is the yield strength of the material divided by the maximum expected
working load.

Figure 4.2: Figure showing the simulation result for the flat grouser under a load of 850 N.

4.1.3 Interpretation and Conclusion

From the results it can be seen that the highest area of stress on the grouser is acting down the
bend line with the maximum stress occurring at the edges of the grouser. This indicates that
this is the region which would fail if a force larger than the maximum expected force was applied.
However, as there is a large factor of safety even if a larger force were to occur, from an event such
as the wheel rolling over a small rock causing a larger localised force, the grouser would still be
able to withstand this.

55
4.2 Chevron Grouser
The chevron grouser is created from a singular piece of metal that is bent first in the centre at an
angle of 20˝ . This creates the 10˝ chevron pattern. It is then bent twice more to create the two
tabs which attach it to the wheel surface.

4.2.1 Simulation Set-up

Two different loading cases are set-up for the chevron grouser. This is because unlike straight
grousers not all of the grouser will make contact with the ground at the same time, when on a flat
surface. The first case tested a distributed load where the 850 N force is split evenly across each of
the three areas on top of the grousers. The two larger areas are both 255.81 mm2 and the smaller
central area is 4.25 mm2 . Therefore in the first case the larger areas had a load of 421.5 N applied
each and the smaller area had 7 N. This set-up can be seen in Figure 4.3.

Figure 4.3: Figure showing the set-up of the distributed load on the chevron grouser before
simulation.

For the second case the load is concentrated on the centre of the grouser in a 2.5 mm wide strip.
This aims to simulate the moment where the grouser first makes contact with the ground. The strip
is projected onto the grouser and results in three segments. The two larger outside areas are both
11.01 mm2 and the smaller central segment is 4.23 mm2 . This can be seen in Figure 4.4. The 850 N
force was then split as 356.5 N on the larger segments and 137 N on the smaller segment.

56
Figure 4.4: Figure showing the set-up of the concentrated load on the chevron grouser before
simulation.

4.2.2 Simulation Results

For the first case the maximum stress was found to be 1.267 ˆ 107 N {m2 at the edge of the lower
bend on either side of the gap at the centre of the grouser. This can be seen in Figure 4.5. A factor
of safety of 28 for Al 7075 or 22 for Al6061-T6 is found.

Figure 4.5: Figure showing the simulation result for the chevron grouser under a distributed load
of 850 N

For the second case the maximum stress is located at the tip of the grouser and is 4.935 ˆ 107 N {m2 .
This can be seen in Figure 4.6. In this case a factor of safety of 7 for Al 7075 or 5.5 for Al6061-T6
is found.

57
Figure 4.6: Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring.

4.2.3 Interpretation and Conclusion

From the results it can be seen that for the first case the the highest stress on this grouser does not
occur down the whole of the bend line. Instead it occurs in a more localised area around the bend
at the centre of the grouser where there is a small portion of the grouser which bridges the gap
between the two halves. As this bridged section is not supported underneath it is expected that
this will lead to higher levels of stress in this area. However the maximum stress is still significantly
lower than the yield stress of the material giving a large factor of safety.
Whilst the grousers factors of safety may appear to be much higher than is required simulations
have been completed assuming an evenly distributed load. In reality as the rover drives over rocky
or uneven terrain the grousers may be subjected to uneven or point loading. In these cases the
forces experienced at those points on the grouser would be many times higher. The second case
tested on the chevron grouser explored this possibility as for the chevron grouser uneven loading
is certain to occur. In the second case the stress is found to be higher than in the first case. This
is because the force is concentrated into a smaller area and stress is simply force divided by area.
Despite the stress being much higher it still remains lower than the yield stress and has a good
factor of safety indicating that if the load were even higher the grouser would still not fail.

4.3 Sawtooth Grouser

The sawtooth grouser is essentially the same as the flat grouser except that a pattern is cut into
the top edge of the grouser to enable the wheel to grip onto hard, rocky surfaces.

4.3.1 Simulation Set-up

The set-up is similar to the flat grouser set up detailed in Section 4.1.1 where the grouser is fixed
along the base and the force is loaded to the top of the grouser edge. However, in this case multiple
points needed to be loaded. As Solidworks takes the sum of each individual point where a force
has been applied the force was evenly split to add up to a total of 850 N, as can be seen in Figure
4.7.

58
Figure 4.7: Image showing the set-up of the sawtooth grouser prior to simulation where multiple
points are used to apply the force.

4.3.2 Simulation Results

It was discovered that when the design was simulated when the teeth had a completely sharp edge
that the stress at the tip was forming a singularity. As in reality the edge would not be perfectly
sharp and would also wear down over time a small fillet of 0.5 mm radius was added to each tooth
edge to model this.

Figure 4.8: Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring.

The maximum stress found was 1.808 ˆ 107 N {m2 which occurred at the tips of the grouser
teeth and can be seen in Figure 4.8. The factor of safety is therefore 20 for Al 7075 or 15 for
Al6061-T6.

4.3.3 Interpretation and Conclusion

From the results it can be seen that while there was still higher area of stress along the bend in
the grouser the highest areas of stress occurred along the tips of the teeth. This is due to the small
contact area that the teeth have with the ground. As previously mentioned when the grouser was
modelled with sharp teeth a singularity was found to occur at the grouser tips as the area was
effectively zero. By remodelling the grouser to have slightly rounded teeth this singularity was
avoided. If it was desired to reduce the stress even further the fillet radius could be increased,
however this would dull the sawtooth edge and make the grouser less effective at gripping onto
hard and rocky terrain. However, it can also be seen that the stress is not high enough to cause
the material to fail at the teeth tips. If the material were to fail at the tips of the teeth the likely
result would be that the teeth would deform, becoming either blunt or bent. While neither of these
deformities would be likely to break the grouser it would reduce the grousers ability to grip onto
hard, rocky terrain therefore it is still undesirable.

59
4.4 Rigid Wheel Spokes
The rigid wheel spokes are used in the rigid wheel design to connect the wheel surface to the wheel
drive motor. A different spokes design is used to the flexible wheel to ensure that the wheel is
rigid.

4.4.1 Simulation Set-up

The rigid wheel spokes simulation configuration had the outer screw holes as fixed parts and it was
loaded with a 850 N force in a downwards direction on the central rim. This can be seen in Figure
4.9.

Figure 4.9: Image showing the set-up of the rigid wheel spokes prior to simulation.

4.4.2 Simulation Results

As seen in Figure 4.10 the rigid wheel spokes had a maximum stress of 1.307 ˆ 108 N {m2 for the
850 N force with the maximum stress occurring at the spokes thinnest point, close to where they
connect with the centre. There is also a secondary region of higher stress which occurs where
the spokes meet with the edge of the rim. Given the yield strength of aluminium 6061-T6 is
2.750 ˆ 108 N {m2 this the rigid wheel spokes results provide a safety factor of 2.1.

60
Figure 4.10: Figure showing the simulation result for the rigid wheel spokes under a load of 850 N.

4.4.3 Interpretation and Conclusion

The maximum stress that is occurring on the rigid wheel spokes is labelled in Figure 4.9 and found
at the outer edge of the spokes. The next highest stress can be found at the spokes thinnest point
where they connect to the centre and around the screw holes on the left and right of the centre.
Even though the factor of safety is relatively lower than the other parts that were individually
tested it is still higher than the minimum safety factor of 2. Therefore the rigid wheel spokes can
withstand the expected maximum force.

4.5 Flexible Wheel Spokes

The flexible wheel spokes are used on the flexible wheel design to connect the springs to the inner
wheel rim.

4.5.1 Simulation Set-up

The flexible wheel spokes simulation configuration had the outer screw holes as fixed parts and it
was loaded with a 850 N force in a downwards direction on the central rim. This can be seen in
Figure 4.11.

61
 Figure 4.11: Image showing the set-up of the flexible wheel spokes prior to simulation.

4.5.2 Simulation Results

It was discovered that when testing the wheel spokes with a 850 N force that the maximum stress
was 1.189 ˆ 107 N {m2 , as seen in Figure 4.12. Given the yield strength of aluminium 6061-T6 is
2.750 ˆ 108 N {m2 this provides a safety factor of 23.

Figure 4.12: Figure showing the simulation result for the flexible wheel spokes under a load of
850 N.

4.5.3 Interpretation and Conclusion

The results show that the wheel spokes for the flexible wheel will withstand the maximum force
that it is expected to experience with a large factor of safety. This is not unexpected due to its
similarity with the wheel spokes on the current SherpaTT wheels. The largest forces are found in
the corners of the spokes that are parallel with the force. As these are the thinnest parts of the
structure it is where the maximum force would be expected.

4.6 Flexible Wheel Rim

The flexible wheel rim is used on the flexible wheel design to connect to the drive shaft that is
attached to the wheel fork. It acts as the connection between the wheel and the wheel drive

62
motor.

4.6.1 Simulation Set-up

The inner wheel rim for the flexible wheel was fixed using the screw holes and loaded with a 450 N
force in a downwards direction on inside surface. This can be seen in Figure 4.13.

Figure 4.13: Image showing the set-up of the flexible wheels inner rim prior to simulation.

4.6.2 Simulation Results

The inner wheel rim for the flexible wheel was found to have a maximum stress of 1.338 ˆ 107 N {m2
when loaded with a 850 N force, seen in Figure 4.14. This provides a factor of safety of 20 if the
wheel rim is made from Al6061-T6.

Figure 4.14: Figure showing the simulation result for the inner wheel rim of the flexible wheel
under a load of 850 N.

4.6.3 Interpretation and Conclusion

From the results it can be seen that the flexible wheel inner rim will be able to withstand the
maximum expected load. It can be see from Figure 4.14 that the maximum stress is occurring at

63
the edge of the main face close to the top most and bottom most three screw holes.

4.7 Full Wheel Assembly

Simulating the full wheel assembly provides a more accurate picture of where the maximum stresses
will be located. This is because the fixtures and forces can be more accurately modelled and the
parts are allowed to interact with each other. As a result the stresses within the model can be
distributed across all the parts. Simulating the full wheel assembly involved a large number of
parts, therefore simplifications were made to the model to reduce the complexity and improve the
simulation time and memory requirements. Screws were removed from the model and the grouser
attachment parts were made as part of the wheel surface.

4.7.1 Simulation Set-up

A static stress test was completed by fixing the lowest point on the wheel and applying a force
inside the central rim, mimicking the action of the drive shaft and the ground. This can be seen in
Figure 4.15. The force used for the simulation was the same 850 N as the other simulations.

Figure 4.15: Image showing the set-up of the rigid wheel prior to simulations

4.7.2 Simulation Results

The full rigid wheel simulation can be seen in Figure 4.16 where the highest stress of 6.692 ˆ 107 N {m2
is found in the lower part of the wheel spokes where it is connecting to the outer rim. Based upon
the individual simulations this is not an unexpected result as stress concentrations were already
noticed to be forming in this area. Another stress concentration can be found on the outer rim
edge opposite the spoke with higher stress. These high stress areas can be seen in more detail
in Figure 4.17. The maximum deflection of this model is approximately 3.2 mm which is a total
deflection of less than 0.85 % of the wheels original diameter. Using the maximum found stress
and assuming the material used is Al6061-T6 this design has a safety factor of 4.1.

64
 Figure 4.16: Figure showing the simulation result for full rigid wheel design.

Figure 4.17: Figure showing the simulation result for full rigid wheel design zoomed into the area
where the maximum stress of 6.692 ˆ 107 N {m2 is occurring.

4.7.3 Interpretation and Conclusion

Simulating the full wheel allowed for a more accurate representation of the forces which will act on
the wheel. It can be seen that the factor of safety found is better than that found in the rigid wheel
spokes simulation in Section 4.4. This is thought to be due to this increased accuracy. However, it
can also be seen that the wheel is deflecting by up to 3.2 mm. Whilst this is not a large proportion
of the total wheel diameter it is still not negligible. Further design improvements could add either a
central supporting ring or another set of wheel spokes to increase the rigidity of the wheel.

4.8 Comparison Wheel Fork

Simulations of the wheel fork remain at a conceptual level. In order to test different wheel fork
configurations several conceptual designs were completed to compare the forces between each of

65
them and to see which had the best stress profile. The initial simulation uses a wheel fork that is
conceptually similar to the current wheel fork on SherpaTT. With a clearance between the top of
9 mm high grousers and the wheel fork being 6 mm.

4.8.1 Simulation Set-up

The simulation is set up by first applying a 850 N force to the top of the model acting as the weight
of the rover above it. A 60 Nm torque is then applied to act as the steering forces. The wheel fork
is then fixed at the bottom by a rod which acts like the drive shaft, which would connect with the
rest of the wheel. This can be seen in Figure 4.18.

Figure 4.18: Image showing the set-up of the conceptual version of the comparison wheel fork prior
to simulations.

4.8.2 Simulation Results

The maximum stress of 3.219 ˆ 109 N {m2 is located at the junction between the drive shaft rod
and the wheel fork. Other higher areas of stress are found at the upper inner bend where the
maximum stress is 1.419 ˆ 108 N {m2 , the outside of the lower bend where the maximum stress is
1.981 ˆ 108 N {m2 and the inside of the lower bend where the maximum stress is 3.195 ˆ 108 N {m2 .
This can be seen in Figure 4.19

66
Figure 4.19: Figure showing the simulation result for the conceptual model of the comparison
wheel fork with a maximum stress value of 3.219 ˆ 109 N {m2 .

By limiting the scale to 2.000 ˆ 108 N {m2 the higher areas of stress can be seen more clearly in
Figure 4.20. Any areas of the model that exceed this smaller scale are shown in black.

Figure 4.20: Figure showing the simulation result for the conceptual model of the comparison
wheel fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow an
easier comparison with the other models.

4.8.3 Interpretation and Conclusion

While the maximum stress can be found at the connection between the drive shaft rod and the
wheel fork this area is not an accurate representation of the actual connection that exists. Therefore
while it should be noted that the stress at this connection should be expected to be higher than
the average stress within the wheel fork it is thought that this stress can be lowered through
improvements on the design. The next high locations of stress are found at the inside upper bend
and on both sides of the lower bend on the wheel fork. As the downwards force is pushing the top
of the wheel fork into the fixed drive shaft it is expected that the inner bends will experience stress
due to this. The addition of the torsion effect that the torque induces causes higher stresses on

67
the outside of the lower bends and increases the stress further down the model. The results from
this model are compared to the other models tested in Sections 4.9, 4.10 and 4.11.

4.9 Double Wheel Fork

The double wheel fork uses a mirrored configuration of the comparison wheel fork and connects
with the wheel on both sides. This configuration was tested in order to see whether distributing the
force down two sides would have a significant improvement on the maximum stress observed.

4.9.1 Simulation Set-up

This simulation is set-up identically to the comparison wheel fork case. There is an 850 N force
and a 60 Nm torque applied to the top of the wheel fork and the drive shaft is fixed at the bottom.
This can be seen in Figure 4.21.

Figure 4.21: Image showing the set-up of the conceptual version of a double wheel fork prior to
simulations

4.9.2 Simulation Results

Similarly to the comparison wheel fork the maximum stress, of 1.452 ˆ 109 N {m2 occurred at the
point where the rod connected with the wheel fork, this can be seen in Figure 4.22. Other areas of
higher stress are found on the outside and inside of the lower bends, the inside of the upper bends
and on either side of the top. The outside of the lower bends have a stress of 5.639 ˆ 107 N {m2
on the left and 6.256 ˆ 107 N {m2 on the right. The inside of the lower bends have a stress of
9.195 ˆ 107 N {m2 on the left and 1.041 ˆ 108 N {m2 on the right. The inside of the upper bends
have a stress of 6.050 ˆ 107 N {m2 on the left and 3.868 ˆ 108 N {m2 on the right. Finally the
bends on either side of the top attachment point have a stress of 3.484 ˆ 107 N {m2 on the left and
3.146 ˆ 108 N {m2 on the right.

68
Figure 4.22: Figure showing the simulation result for the conceptual model of a double wheel fork
with the scale showing the maximum stress of 1.452 ˆ 109 N {m2 .

By limiting the scale to 2.000 ˆ 108 N {m2 the higher areas of stress can be seen more clearly in
Figure 4.23. Any areas of the model that exceed this smaller scale are shown in black.

Figure 4.23: Figure showing the simulation result for the conceptual model of a double wheel
fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow an easier
comparison with the other models.

4.9.3 Interpretation and Conclusion

The areas of high stress found on the double wheel fork were in similar locations to those found on
the comparison wheel fork. As the simulation was set-up in the same way this is to be expected.
There was a new area of higher stress found around the top where the force is pushing down. It is
thought that this area did not appear in the comparison wheel fork results as the stress was overall
higher. As the torque is twisting the wheel fork in a clockwise direction the forces are not even on
each side of the wheel fork. The force is higher on the left side in the top of the model and on the
right side on the right of the model. This is due to the drive shaft being fixed in between the lower
ends causing torsion in the model.
By splitting the force down both sides of the double wheel fork the maximum stress is significantly

69
less than the comparison wheel fork, shown in Section 4.8. The maximum stress found is 2.22
times smaller that the comparison wheel fork. Other comparable stresses are at least 3.07 times
smaller on the inside of the lower bend, at least 3.16 times smaller on the outside of the lower bend
and at least 2.35 times smaller on the inside of the upper bend. This has the potential to allow
for a thinner wheel fork to be used. However, there is a drawback to this design in terms of rock
entanglement. As the wheel fork now extends down both sides of the wheel there is a potential for
rock entanglements on both sides of the wheel instead of just one side. This risk can be partially
mitigated by decreasing the thickness of the wheel fork but the thickness cannot be reduced too
much otherwise the stresses would be too great and the wheel fork would fail.

4.10 Taller Wheel Fork

As one potential method of reducing the chances of rock entanglement is to increase the clearance
between the grousers and the wheel fork a conceptual model that is taller than the comparison
model has been created. The clearance between the top of 9 mm high grousers and the wheel fork
is 31 mm for this case.

4.10.1 Simulation Set-up

This simulation is set-up identically to the previous cases. There is an 850 N force and a 60 Nm
torque applied to the top of the wheel fork and the drive shaft is fixed at the bottom. This can be
seen in Figure 4.24.

Figure 4.24: Image showing the set-up of the conceptual version of the taller wheel fork prior to
simulations

4.10.2 Simulation Results

The highest stress, of 3.994 ˆ 109 N {m2 , is again found to be acting at the point where the drive
shaft joins the base of the wheel fork. This is followed by three other areas of higher stress which
are found on the outside of the lower bend, the inside of the lower bend and the inside of the
upper bend. The outside of the lower bend has a maximum stress of 2.223 ˆ 108 N {m2 , the inside
of the lower bend has a maximum stress of 3.505 ˆ 108 N {m2 and the inside of the upper bend has
a maximum stress of 1.505 ˆ 108 N {m2 .

70
Figure 4.25: Figure showing the simulation result for the conceptual model of the taller wheel fork
with a maximum stress value of 3.994 ˆ 109 N {m2 .

By limiting the scale to 2.000 ˆ 108 N {m2 the higher areas of stress can be seen more clearly in
Figure 4.29. Any areas of the model that exceed this smaller scale are shown in black.

Figure 4.26: Figure showing the simulation result for the conceptual model of the taller wheel
fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow an easier
comparison with the other models.

4.10.3 Interpretation and Conclusion

Comparing the results of this study to the study on the comparison wheel fork, Section 4.8, it can
be seen that the distribution of the stress is mostly the same. Since the only difference between
these two models is an increase in the vertical height the only expected change was for an increase
in the stress magnitude. By looking at the stresses found on each model it can be seen that the

71
maximum stress is 1.24 times higher than the comparison wheel fork. Comparing the other areas
of higher stress to the comparison wheel fork it can be seen that the outside of the lower bend is
1.12 times higher, the inside of the lower bend is 1.10 times higher and the inside of the upper
bend is 1.06 times higher. Therefore if the wheel fork is made longer in height the inner structure
of the wheel fork must provide a higher level of support to the identified areas to ensure that it
does not fail.

4.11 Slanted Wheel Fork

Another potential method of reducing the chances of rock entanglement is to use a slanted wheel
fork. Using a slanted wheel fork allows the wheel forks effective thickness to be lower where it is
closer to the ground and more likely to encounter rocks and allows it to be thicker higher up. This
design can be used with an angle that puts the end of the wheel fork inside the wheel at the lower
edge and slants upwards until it is clear of the wheel outer rim on the top edge. To create the slant
the wheel fork’s vertical bar has been changed from a 90˝ angle with the upper horizontal bar to
a 70˝ angle.

4.11.1 Simulation Set-up

This simulation is set-up identically to the previous cases.There is an 850 N force and a 60 Nm
torque applied to the top of the wheel fork and the drive shaft is fixed at the bottom. This can be
seen in Figure 4.27.

Figure 4.27: Image showing the set-up of the conceptual version of a slanted wheel fork prior to
simulations.

4.11.2 Simulation Results

This model had a similar stress hot spot as the other models at the point where the drive shaft
joins the base of the wheel fork with a maximum stress of 3.058 ˆ 109 N {m2 , which can be seen
in Figure 4.28. It then has three other stress concentrations, one on the inside of the lower bend,
one on the outside of the lower bend and one on the inside of the upper bend. The outside of
the lower bend has a maximum stress of 1.984 ˆ 108 N {m2 , the inside of the lower bend has a
maximum stress of 3.238 ˆ 108 N {m2 and the inside of the upper bend has a maximum stress of
1.983 ˆ 108 N {m2 .

72
Figure 4.28: Figure showing the simulation result for the conceptual model of the slanted wheel
fork with a maximum stress value of 3.058 ˆ 109 N {m2 .

By limiting the scale to 2.000 ˆ 108 N {m2 the higher areas of stress can be seen more clearly in
Figure 4.29. Any areas of the model that exceed this smaller scale are shown in black.

Figure 4.29: Figure showing the simulation result for the conceptual model of the slanted wheel
fork with the scale going up to a maximum stress value of 2.000 ˆ 108 N {m2 to allow an easier
comparison with the other models.

4.11.3 Interpretation and Conclusion

From the results it can be seen that the stress concentrations are occurring in the same locations as
they are on the comparison wheel fork, by comparing Figure 4.29 and Figure 4.20. As the force and
the torque are applied in the same way this is expected. Overall the slanted wheel fork experiences
lower stress than the comparison wheel fork. With the maximum stress on the comparison wheel
fork being 1.05 times higher than the slanted wheel fork. Comparing the other areas of higher
stress to the comparison wheel fork it can be seen that the outside of the lower bend is almost
negligibly higher, the inside of the lower bend is 1.01 times higher and the inside of the upper bend
is 1.40 times higher. From this it can be seen that the only area that is experiencing a significant
increase in stress is the inside of the upper bend. This is because the distance from the point where

73
the force is loaded to the upper bend is larger than on the comparison wheel fork to allow for the
70˝ slant angle. As moment is equal to force multiplied by distance it is expected for the force to
increase in this way. This increase in stress could be negated if the wheel fork connected to the
inner rim of the wheel inside the wheel, instead of connecting to the outside of the wheel. Doing
this would allow the distance between the force and the bend to be decreased thus decreasing the
moment force.

4.12 Summary
The first parts tested were the different grouser designs. These were modelled as sheet metal
parts using a distributed force in all but one case. The results from this showed that the grousers
would withstand the maximum expected stress with a large margin of safety. This is an important
outcome as when wheels are driving over rocky or uneven terrain the grousers will not experience
a distributed load as certain areas will have a higher loading. As the chevron grouser will regularly
experience a more concentrated load this was also tested with the results showing that the grouser
would still not fail. This means that the grousers which have been developed from the preliminary
designs are robust enough to be used on SherpaTT.
Next, the inner parts of the two wheel types were tested. The spokes for both the flexible and rigid
wheel and the inner rim of the flexible wheel were shown to be able to withstand the weight of
the rover. While the rigid wheel had a smaller factor of safety it is still high enough that it should
not fail. If the payload capacity of SherpaTT was to be increased the rigid wheel spokes could be
made thicker in order to withstand an increased load.
The full rigid wheel assembly was then tested. Whilst the grousers were included the screws were
not in order to decrease the time and memory requirements of the simulation. The factor of safety
found was in fact higher than the factor of safety found when testing the wheel spokes alone. This
could be due to the full assembly better modelling the interactions between parts. The flexible
wheel was not tested as a full assembly due to restrictions within Solidworks modelling. However,
as the springs and attachment pieces have not been changed on the flexible wheel design it can be
assumed that these pieces will work as they are currently in use on SherpaTT.
Finally the different wheel fork concepts were tested. Comparing the results from the double
wheel fork, in Section 4.9, taller wheel fork, in Section 4.10, and the slanted wheel fork, in Section
4.11, with the comparison wheel fork, in Section 4.8 provides a view on some of the benefits and
drawbacks of each concept. Overall the double wheel fork has the lowest stress, improving on the
comparison wheel fork, while the taller wheel fork has the highest stress of all wheel forks tested.
Adding a slant increases the stress in the upper part of the wheel fork more than any other tested
concept, however this could be mitigated if the wheel fork can be placed partially inside the wheel.
If either the taller or slanted cases the wheel fork were used on the current SherpaTT wheel fork the
inner structure would need to be improved upon to ensure it could withstand the higher stresses.
Finally whilst each of these concepts has been tested individually it would be possible to combine
the concepts. For example combining the taller and slanted concepts to reduce the chances of rock
entanglement in two ways simultaneously, first by increasing the clearance allowing any rocks that
do become stuck between the grousers space to pass freely and second by making it more difficult
for the wheel fork to sit on top of rocks and more likely to slip off them.

74
Chapter 5

Final Results

As discussed in Chapter 3 the optimum design for the wheels varies depending on what kind of
terrain is being targeted. The requirements are also different depending on whether a flexible or
rigid wheel is used. Therefore results are presented for four different scenarios, a rigid wheel on
coarse sand, a rigid wheel on soft sand, a flexible wheel on coarse sand and a flexible wheel on soft
sand. The final CAD designs for each wheel is shown and described and the chapter ends with
recommendations on material selection based upon the durability investigation.

5.1 Final CAD

Whilst optimised designs for different targeted terrain are given there are several features that
appear in all the designs. The most important of these features is regarding the chances of rock
entanglements with the new designs.
Overall the grousers are formed to reduce the chances of any rock entanglements in three main
ways. Firstly the grousers are now placed at 90˝ angles normal to their contact point on the wheel,
instead of the current 45˝ angle. By increasing this angle it will act to prevent rocks from becoming
wedged in between a grouser and the wheel surface. Secondly the grousers are no longer paired
together, with the same sized gap existing between all grousers. Previously the gap between each
grouser was 9.4˝ and 8.6˝ with an arc length of 32 mm and 29 mm. Splitting up the grousers makes
the spacing 12.9˝ , or an arc length of 44 mm, in the case of 28 grousers, or 18˝ , or an arc length of
61 mm, in the case of 20 grousers, apart increasing the minimum distance between grousers by up
to 9.4˝ or or an arc length of 32 mm. By increasing the separation smaller, lighter rocks should be
prevented from becoming wedged in between the grousers. If larger, heavier rocks were to become
stuck they are less likely to remain stuck as they weigh more. Finally, by ensuring that the grouser
height has not been too significantly extended compared to the current SherpaTT grousers it also
ensures that the overall pocket created by the grousers and the wheels is not much deeper, which
could invite larger rocks to become trapped.

5.1.1 Rigid Wheel on Coarse Sand

This design is optimised for the ES-3 to ES-4 category, from Table 2.1. In this regolith category the
rover will drive mostly across terrain with a higher coefficient of friction and cohesion, experiencing
less sinkage. Therefore the design will use a higher number of shorter grousers. Based upon the
calculations completed in Section 3.3.2, the number of grousers used will increase from 10 double
grousers to 28 single grousers and the height will be 9 mm providing a spacing of 12.9˝ . As this
wheel is targeting harder terrain the grousers will use a sawtooth edge in order to aid the rover
when driving over rockier terrain. The wheel is also less likely to experience high levels of slip
or sinkage on harder ground, therefore, since the greatest tractive efficiency comes from straight
grousers this is what will be utilised on this wheel. The final wheel design can be seen in Figure
5.1.

75
 Figure 5.1: Figure showing the final CAD for a rigid wheel on coarse sand.

The simulation results for this wheel are detailed in Section 4.3 for the grousers, Section 4.4 for
the inner wheel spokes and Section 4.7 for the full wheel simulation. These results show that this
wheel can withstand an 850 N force with a safety factor of 4.1 when using Al6061-T6 for the inner
structure and wheel surface and Al7075 for the grousers.
The recommended combination with the wheel fork concepts is to use a slanted wheel fork and a
taller wheel fork. This is because the end of the wheel fork could have the potential to sit within
the wheel’s width due to the fact that there is only one set of spokes leaving the other side of the
wheel open. Using a taller wheel fork will increase the clearance between the top of the grousers
and the wheel fork which will allow any small rocks which do become stuck to pass freely through
this gap. The double wheel fork concept would not work as well with this design as a double fork
would require an attachment point on both sides of the wheel which would be more difficult as the
spokes are only attached on one side of the wheel. Using a double fork would be possible but it
would require some redesign of the inner wheel parts.

5.1.2 Rigid Wheel on Soft Sand

This design is optimised for the ES-2 to ES-3 category, from Table 2.1. In this regolith category the
rover will drive mostly across terrain with a lower coefficient of friction and cohesion, experiencing
more sinkage. Therefore the design will use a lower number of taller grousers compared to the
rigid wheel on coarse sand scenario. Based upon the calculations completed in Section 3.3.2, there
will be 20 single grousers at a height of 15 mm. The grousers will have a flat edge as there will be
greatly reduced need for the ability to get a grip on extremely hard surfaces in this scenario. A
chevron pattern will be used to aid the wheel in both avoiding side slip and to help prevent sinkage
that could result from the grousers digging the wheel in deeper as chevron grousers are less prone
to digging compared to straight grousers. The final wheel design can be seen in Figure 5.2.

76
 Figure 5.2: Figure showing the final CAD design for the rigid wheel on soft sand.

The simulation results for this wheel are detailed in Section 4.2 for the grousers and Section 4.4 for
the inner wheel spokes. These results show that this wheel can withstand an 850 N force.
The recommended combination with the wheel fork concepts is the same as in Section 5.1.1, to use
a slanted wheel fork and a taller wheel fork. This is for the same reasons as detailed above.

5.1.3 Flexible Wheel on Coarse Sand

This design is optimised for the ES-2 to ES-3 category, from Table 2.1. In this regolith category the
rover will drive mostly across terrain with a lower coefficient of friction and cohesion, experiencing
more sinkage. As a flexible wheel is overall more suitable for soft, deformable soil as the amount
of sinkage experienced by a flexible wheel is reduced compared to a rigid wheel of the same
dimensions. Therefore either smaller or fewer grousers are required compared to the rigid wheel
case. Additionally increasing the number of grousers will have a secondary effect of decreasing the
flexibility of the wheel. Based upon the calculations completed in Section 3.3.2, there will be 20
single grousers at a height of 9 mm.
Both of the flexible wheel designs use straight grousers despite the fact that straight grousers offer
less side slip resistance and are more prone to sinkage due to their digging motion. This is because
utilising any kind of grouser pattern would have a negative effect on the wheels ability to deform
thus limiting the flexibility of the wheel. The final wheel design can be seen in Figure 5.3.

77
 Figure 5.3: Figure showing final CAD for the flexible wheel on hard ground.

The simulation results for this wheel are detailed in Section 4.3 for the grousers, Section 4.5 for the
inner wheel spokes and Section 4.6 for the inner rim. Based upon these results and the unchanged
parts of the current SherpaTT wheel design this wheel will be able to withstand a force of at least
850 N.
The recommended combination with the wheel fork concepts is to use a taller wheel fork to increase
the clearance between the top of the grousers and the wheel fork thereby allowing any small rocks
which do become stuck to safely travel through the gap. There is also the possibility to combine
this with either a double fork or a slanted fork. Using a double fork would significantly decrease
the stresses within the wheel fork, however it would introduce the possibility of rock entanglements
on both sides of the wheel. Using a slanted fork would not decrease the overall width of the wheel
fork in this case as it would not be able to sit within the wheel at all. However, using a slanted
fork would encourage the rover to slip off rocks which it may pose entanglement threats instead of
getting stuck on them.

5.1.4 Flexible Wheel on Soft Sand

The final wheel design can be seen in Figure 5.4 and is targeting the ES-2 to ES-1 soil types. Based
upon the calculations completed in Section 3.3.2, this design utilises 20 single grousers at a height
of 15 mm. This increased height will aid the wheel in removing the sand in front of the wheel
before the wheels outer rim makes contact with the ground, thus reducing the rolling resistance
and increasing the wheel efficiency. As this terrain will be very soft there is no need for the use of
sawtooth shaped grousers as the wheel is unlikely to have the need to have an improved ability to
grip onto hard, rocky terrain.

78
Figure 5.4: Figure showing the simulation result for the toothed grouser under a load of 450 N.
The view is from below to show the area where the maximum stress is occurring.

The simulation results for this wheel are detailed in Section 4.1 for the grousers, Section 4.5 for the
inner wheel spokes and Section 4.6 for the inner rim. Based upon these results and the unchanged
parts of the current SherpaTT wheel design this wheel will be able to withstand a force of at least
850 N.
The recommended combination with the wheel fork concepts is the same as in Section 5.1.3, to use
a taller wheel fork and potentially combine it with a double or slanted fork. This is for the same
reasons as detailed above.

5.2 Durability Investigation Results

Based upon the information gathered during the durability investigation this thesis makes several
recommendations with regards to material selection for the grousers.
At different points of the SherpaTT’s project lifetime the budget and constraints can be expected to
differ. For example at the current stage of development SherpaTT is largely built and maintained in
house using the machinery and expertise available. Therefore at this stage this thesis recommends
that for the grousers anodised Al 6061-T6 is used due to its reasonable hardness, which will slow
down wear, its ease of machinability, which will aid in manufacture and maintenance ease and
the ability for anodising which will increase the surface hardness and also protect against wear.
Another material that could have some potential is Al 2014-T6, however, despite having better
strength and hardness characteristics the 2000 series is susceptible to corrosion which is highly
likely to become a problem for SherpaTT in its current work role. One potential workaround to
this corrosion issue would be to use a coating, however, if this coating was to wear away the metal
would then once again be susceptible. When the project reaches a later stage where it is ready to
be used in space it would be recommended to make the grousers from anodised Al 7000 class of
aluminium. The Al 7075 class provides a much higher strength and hardness level than the 6060
class meaning that the grousers would last significantly longer compared to the current grousers.
However, this comes coupled with added costs and more difficult manufacturing methods. Two
potential options could be Al 7075 or Al 7475.

79
For the wheel spokes and surface the material recommendation also depends on the SherpaTT’s
current development status. At the current point the Al 6000 class Al6061-T6 would be most
recommended due to its good strength characteristics, corrosion resistance and its ease of extrusion
and forming. In situations where it may not be hard enough anodising can be used to give a small
boost to the hardness. Once the project moves into the later stages it may wish to look into
using a higher grade of aluminium, such as from the 7000 class or when looking at more critical
components at titanium. 5Al-2.5Sn-Ti
Whilst titanium has excellent properties, as seen in Table 3.4, it is significantly more expensive
than aluminium. Therefore, if aluminium can sufficiently do the job then it is generally preferable
to use it over titanium. There may be some critical parts or components where aluminium can
not meet titanium’s strength to weight ratio in which case it is safer to use titanium. Steel
exhibits similar properties to the aluminium 7000 class but has nearly three times the density.
This makes it generally unsuitable for space missions as high grade aluminium can replace it in
most situations.

5.3 Summary
The design of each wheel was selected based upon the results from the research into what is
currently being used on rover wheels and into the state of the art and combining this with the
results from the analytical study into the grousers and the simulations that were completed to
test the design. The new designs both reduce the chances of rock entanglements with the wheel
and provide optimal traction under four different conditions. Whilst each wheel is optimised for
a certain soil type all of the wheels have the ability to cross any of the soil types, however the
efficiency of the wheel would be reduced which would impact on the distance per day that the
rover is able to drive. If a mission were to be heading to an area that is likely to see an even
amount of all the soil types then the recommendation would be to select a wheel that is optimised
for either ES-3, which if an extra category for solid ground was included after ES-4, would put it
in the centre of the scale. In this case either the rigid wheel for soft sand, Section 5.1.2, or the
flexible wheel for coarse sand, Section 5.1.3, would both be reasonable choices. To choose either of
the other two wheels would lead to a much larger degradation in efficiency at the opposite end of
the scale.
Based upon the conceptual studies into the different wheel forks recommendations have been made
on which designs would suit each wheel. In all cases it was recommended to increase the height of
the wheel fork to allow for a greater clearance between the top of the grousers and the wheel fork to
reduce the chances of rock entanglement. The use of a slanted wheel fork was also recommended to
encourage the wheel fork to slide off any rocks that it may start to become entangled with instead
of becoming stuck.
Finally recommendations about which materials would be the most suitable were made with
potential options being Al6061-T6 for use in the main wheel structure and Al7075 for use in
the grousers. It was noted that the materials chosen may vary within the projects lifetime as the
priorities shift. Whilst the rover is still in a testing or proof of concept stage it would be better to
use materials that are easier to work with to simplify construction and maintenance, such as the
6000 or 2000 aluminium classes. However, when the rover enters its final stages of development
and the lifetime of a part is more important even if parts can no longer be manufactured in house
then materials that have higher strength or hardness values would be better, such as the 7000
aluminium class.

80
Chapter 6

Discussion

6.1 Summary
The SherpaTT rover wheels and lower leg structure were investigated to discover where current
problems lie with regards to rock entanglement issues. It was found that there were a number
of rock entanglements occurring during the field trials which was undesirable in a rover that is
targeting more challenging terrain. These rock entanglements could all be classified as either
grouser entanglements or wheel fork entanglements.
Past, current and future rovers were then studied to see what had and hadn’t worked and what
potential new ideas existed. A focus was made on the past, current and future Mars rovers as
they are designed with the Martian terrain in mind. From this study it was found that most
of the wheels that have been used on Mars in the past have had some issues with their wheels,
either becoming stuck or showing signs of early wear. Even though these issues were not good for
the missions they do provide a good learning experience in what does and does not work. The
past missions also showed that despite excellent preparation and significant studies of the Martian
terrain before the mission it is still very difficult to predict what type of regolith the rover will have
to traverse over. This is especially true in areas where a thin crust covers soft sand beneath.
Research was completed into the Martian environment, looking at the terrain, thermal and weather
conditions as well as potential mission locations. This revealed that the Martian regolith can be
categorised into different types depending on its granular size. As ESA, among many others, have
their own set synthetic recipe for these regolith types it was possible to use data about the synthetic
regolith types for use in an analytical study. Studying the thermal environment showed that the
Martian temperature changes are linked with the thermal inertia of the surrounding terrain. A
higher thermal inertia is better as it means the area will retain the heat from the day for longer
resulting in smaller daily temperature changes. As thermal inertia is dependent largely on the
surrounding terrain it can also be an indicator of what terrain is present in an area, where a high
thermal inertia indicates there is likely to be a higher proportion of rocks or exposed bedrock and a
low thermal inertia indicates there is likely to be a large amount of dust coverage. This knowledge
can be used to gain a better understanding of what the terrain may be like at various different
potential mission locations on Mars. The missions locations were examined based upon scientific
merit, viable landing possibility, likely terrain conditions and likely mission constraints. Initially
five were investigated but this was narrowed to one, Nilosyrtis Mensae. Nilosyrtis Mensae offers a
strong potential for good scientific outcomes whilst also broadly following the constraints as used
by the MER and MSL rovers. The terrain at Nilosyrtis Mensae is likely to have some dust coverage
based on the thermal inertia of the area, however it is within the thermal inertia range of areas
rovers have previously visited on Mars. High resolution imagery appears to confirm this as the area
is largely free from ripples which would indicate a sand dune area. To confirm likely soil conditions
further high resolution imagery and thermal inertia measurements would be required.
A review was carried out to look into the analytical side of the design. Work by Bekker on a rigid
wheel on soft ground, which was later elaborated on by Reece and Wong, was found and utilised
to get a basic understanding of wheel terramechanics. Bekker’s method used the coefficient of

81
cohesion and the coefficient of friction of the soil to describe the wheel’s interaction with the soil.
A major drawback of this is that the soil must be tested and measured first, in order to get these
coefficients. Since there has currently been no return of Martian regolith to Earth there is no way
to test this in a lab. However efforts have been made with experiments on the MER rovers which
has allowed for a reasonably accurate Martian soil simulant to be created. Using the data from the
ESA Martian soil simulant it was then possible to calculate how much the wheel would sink into
this soil. As the two coefficients of cohesion and friction are dependent on many factors within
the soil, including the soil density, data was used that included various densities of the simulants.
This study using Bekker theory was then combined with later work by Moreland which researched
how the grousers were interacting with the soil in order to try and produce design guidelines. This
work suggests that by ensuring the grousers are spaced so that a grouser sweeps away the soil that
is in front of the wheel before the wheels outer rim can touch the soil that the tractive efficiency of
the wheel will be optimised. By combining the two it was possible to see how the grouser height
and spacing interacted with each other and how changing other parameters such as the soil type or
even the rover weight or wheel size could impact upon both the number of grousers recommended
and the sinkage of the wheel.
The design stage began by looking broadly at what aspects of the wheel and wheel fork could
be changed so the the chances of rock entanglement were reduced. These initial parameters were
mapped out and evaluated for their strengths and weaknesses. An early decision was made that
the angle of the grousers should be changed in order to prevent rocks from becoming wedged in
the gap between the grousers and the wheel surface. This was backed up with the knowledge that
all but one of the past, present and future rovers had used grousers at 90˝ to the wheel, with the
exception being Sojourners spikes, therefore this was a tried and tested set-up that worked. In
order to decide what other changes should take place a more analytical approach was required and
Bekker’s method of terramechanics and Moreland’s grouser equation were used. These produced
recommendations on the height and number of grousers. However, as the height and number are
interdependent a range of values was produced going from a large number of very short grouser
to a small number of very tall grousers. As the relationship is exponential at each end of the
curve there are extreme values meaning that an optimum solution is found in the centre range.
The height and number decision were made based three things. The current grouser set up that
SherpaTT currently has, the aim to reduce rock entanglements by discouraging rocks from sticking
in the gaps between grousers and knowledge of rigid and flexible wheel characteristics.
Once the preliminary designs had been created in CAD form considerations were taken as to which
grousers would be easier to manufacture and which would be harder. Creating a straight grouser
simply requires the folding and cutting to size of sheet metal however creating something like the
wave pattern to be manufactured is much trickier. Ultimately it was decided that unless these
complex patterns were to be welded or machined onto the wheel surface it would be better to
use the less complex patterns as they would be able to be more easily manufactured and thus
more robust in their attachment to the wheel. Additionally there are more studies on the straight,
chevron and slanted patterns which provide a higher level of confidence on how well the wheel
should perform. The final patterns were chosen based upon the expected tractive efficiency on the
terrain, resistance to side-slip and sinkage and whether or not the design would inhibit a flexible
wheels ability to be flexible. Each design was then further iterated in CAD with simulations
completed to ensure that the design would be able to withstand the force from the rover. Also
investigated in CAD simulations were several different wheel fork concepts. By using a comparison
wheel fork that is a conceptual model of the current wheel fork and changing one aspect to create
three new cases the performance of each change could be evaluated. Results from this study showed
where on the wheel fork the higher areas of stress would be found and in particular which areas
showed an increase in stress following the changes.
During this period a material investigation was also completed to find out which materials are
suitable for use in a Martian mission. As many materials become very brittle in extreme cold it
was particularly important to ensure that this was not the case for the recommended materials.
Consideration was also given with regard to the materials workability which impacts upon cost
and ease of maintenance.
Lastly the final wheel designs were presented with consideration given to the potential wheel fork
changes that could be coupled with them. The final designs take the design choices made in

82
the previous stages and combine them onto both a rigid and a flexible wheel. These results are
discussed and recommendations made as to which configuration is most suitable for each different
scenario examined.

6.2 Conclusion
Using the Bekker model of terramechanics and combining it with Moreland’s grouser equation has
provided guidelines on suitable grouser heights depending on the grouser number and on the soil
type. From this it was concluded that the current set-up on SherpaTT, with its 10 double grousers,
was spaced too far apart for their height. As the height was already at 9 mm the spacing between
the grousers was decreased. By decreasing the spacing the grousers will be able to sweep away
soil that is in front of the wheel before the wheel surface can make contact with the ground. This
reduces the rolling resistance on the wheel therefore increasing tractive efficiency. Another solution
was to increase the height of the grousers, however this comes with new problems. Taller grousers
will not be able the withstand the same amount of force before failing as shorter grousers and
taller grousers decrease the clearance between the top of the grousers and the wheel fork further.
Therefore the grouser height was only increased for the softer soil conditions where maintaining
the same grouser height would result in a large increase in grouser number and where the forces
on the grousers would be less as they sink into the softer soil. Additionally to reduce the chances
of rock entanglements the grousers were changed from having an angle of 45˝ to an angle of 90˝ .
By doing this the chances of a rock becoming stuck in between the grouser and the wheel surface
is reduced. Finally the grousers were split so that they were single, not double, grousers. This
change occurred as even though the grousers were overall spaced too far apart the two grousers
on each double part were spaced too close together. Having this close spacing meant that it was
easier for rocks to become stuck inside the double grousers. By separating out the grousers this
rock entanglement method is reduced.
All the newly designed parts were tested using Solidworks simulation static case to check for where
the high stress concentrations were occurring. The designs were then iterated to ensure that they
first, were at minimum a safety factor of 2 underneath the yield stress and that second, they were
optimised where possible to reduce their total mass. Several singularities were found in many of the
components during simulations. These were resolved using small fillets to better represent sharp
edges and ensuring that the most appropriate body type was being used on sheet metal pieces.
Testing was also attempted on the flexible wheel, however despite many attempts and significant
simplification a successful simulation was not achieved. From the testing on the wheel parts it
can be concluded that all the parts will be able to withstand the forces that they will experience
from the weight of the rover. As the Martian gravity is much less than Earth’s the factor of safety
achieved considering a Martian environment is even greater.
Whilst the flexible wheel inner section was mostly left the same as previous versions the rigid
wheel inner was developed. As part of this development the central wheel hub was brought
11.75 mm closer to the edge of the wheel. Doing this allows for the torque sensor to fit a little
more comfortably inside the wheel and leaves a smaller amount of overhang in the wheel fork.
This therefore reduces the chances of a wheel fork entanglement as it allows for the wheel fork to
connect closer to the wheel.
From the wheel fork conceptual studies it can be seen that the alternative designs are feasible
in terms of maximum stress concentrations. However, as these designs remain at a conceptual
level further studies would be required. When looking at the results from a rock entanglement
perspective there are several improvements that can be achieved. The slanted wheel fork eliminates
most of the wheel fork bench that the wheel fork could prop itself up on and would encourage the
wheel and leg to slide off any rock that might get in the way. This reduces the chances of rock
entanglement with the wheel fork. However, the slanted wheel fork has an increased level of stress
on the inside of the upper bend. Whilst this could be partially mitigated by attaching the wheel
fork at a point inside of the wheel, instead of to the outside, care must still be take to ensure
the wheel fork could withstand the stress. The double fork has a huge impact on reducing stress
within the wheel fork. It may therefore be possible to reduce the width of the prongs on the wheel
fork which would make each side sit closer to the wheel. Having less of the wheel fork sticking

83
out from the wheel reduces the chances of a rock entanglement with the wheel fork. However,
even if the width was decreased there would still now be a wheel fork on both sides of the wheel
meaning that either side of the wheel could potentially become entangled. The taller wheel fork
increases the clearance between the top of the grousers and the wheel fork. This allows any rocks
that do become stuck in the grousers to pass safely through thus reducing the chances of rock
entanglement with the grousers and wheel fork. However, as the taller wheel fork has higher stress
levels than a shorter wheel fork care must be taken to ensure the inner structure of the wheel fork
can withstand the forces.
Investigations into the various materials which could be used for construction introduced the idea of
using a harder metal or an anodised metal for the grousers to increase the hardness and resistance
to wear. Care must be taken when anodising that the hole sizes for screws and other measurements
which require a high level of precision take into account the extra thickness that will be added by
anodising. For the rest of the wheel where hardness is not such an important trait materials were
considered in terms of their strength. As stronger and harder materials are generally harder to
work with the recommendations for different stages of the project are different. With aluminium
from the 6000 class such as Al6061-T6 being recommended for all parts at the current stage,
anodising the grousers to increase the hardness further, and aluminium from the 7000 class such
as Al7075-T6 begin used at least on the grousers when SherpaTT is sent to Mars. Additionally it
is recommended that were there are critical components which could result in a catastrophic wheel
failure that titanium, such as 5Al-2.5Sn-Ti, is used to provide a larger factor of safety.
Depending on the ultimate location of the mission several different wheel designs have been
proposed. All four of the proposed wheels would work in any location, however, they are each
more efficient over more specific terrain types. Therefore if the terrain planned for driving over
appears to consist more of one terrain type then a wheel optimised more for this terrain will provide
a higher level of driving efficiency over the course of the mission. If the terrain at a mission site
is a variety of different types then it is most suitable to go for one of the wheel designs which is
aimed at the middle of the soil types so that it is not being pushed to work too far away from the
soil it was optimised for.
Overall it can be concluded that wheel design is very interdependent. As the wheel size changes
so does its traction and its sinkage, which then affects the grouser layout which in turn affects
the wheels traction and its response to sinkage. What kind of wheel is required depends on the
mission constraints, particularly constraints on size, weight and cost, the terrain in which the
wheel is expected to operate in, whether or not the wheel is maintainable and serviceable during
its lifetime and the type of vehicle the wheel is supporting. Wheel terramechanics is still not
fully understood with the current models requiring many assumptions in order to work. There is
also no analytical model existing yet for a non-pneumatic flexible wheel on soft ground meaning
that most work comes through trial, test and error. The new wheel designs reduce the chances of
rock entanglement by, changing the angle of the grousers to 90˝ , increasing the minimum spacing
between the grousers on all four wheels and without increasing the height on two of them, by
recommending that the clearance between the wheel fork and the wheel is increased and a slant
added to the wheel fork.

6.3 Improvements and Future Work

A wide array of future work exists within this project. The first potential area would be to focus
in on the grousers in a much more analytical way. Taking the Bekker method further and utilising
the work done by Reece, Wong and others to model the wheel and its grousers both in the static
and dynamic state. Leading on from that is also the potential to carry out DEM simulations
which would allow a wheel to be modelled in 3D which would give a fairly accurate prediction of
performance. Finally now that the new grousers have been designed the next step would be to test
them. This could be as simple as just testing the set-up’s as they are proposed in this thesis or it
could be undertaken as a study within itself and test different combinations of grouser numbers,
grouser heights and grouser patterns on both the rigid and the flexible wheel and check if reality
matches with the design. It would also be interesting to carry out some wear tests using grousers
of different materials and different coatings to see which one lasts the longest and if it is what is

84
expected.
The second potential area would be to focus on the wheel fork, taking the results from the
conceptual designs that are presented here and implementing them into a full redesign of the
wheel fork. Once fully redesigned it should be tested first within a computer simulation to ensure
the maximum stress is not excessive before it can be build and tested with the rover. Having
the conceptual designs already completed will provide a good indication on where stresses can be
expected to be higher allowing for the design to reinforce these areas early.
A third potential area would be to look in more detail at the material choices for the wheel
and possibly the entire structure. Ensuring that the materials used are suitable for use in space
and that they won’t exhibit unwanted behaviours during the extreme temperatures or pressures
found during the launch, transfer, landing or operational lifetime. Investigating whether any parts
can have their mass reduced either through using a lighter material, through milling out excess
structure or using a manufacturing method that results in fewer extra parts. For example instead
of then attaching two parts together using screws to use a single piece of metal and mill it to result
in only one part and no screws.
The final potential area would be to either build or 3D print the current wheel designs and to test
their performance. Weaknesses in the design can then be identified and improved upon.

85
Bibliography

[1] Michael Manga, Ameeta Patel, Josef Dufek, and Edwin S. Kite. Wet surface and dense
atmosphere on early mars suggested by the bomb sag at home plate, mars. Geophysical
Research Letters, 39(1), 2012.
[2] M. Grott, A. Morschhauser, D. Breuer, and E. Hauber. Volcanic outgassing of co2 and h2o
on mars. Earth and Planetary Science Letters, 308(3):391 – 400, 2011.
[3] Gaetano Di Achille and Brian Hynek. Ancient ocean on mars supported by global distribution
of deltas and valleys. Nature Geoscience, 3:459–463, 06 2010.
[4] Igor Levchenko, Shuyan Xu, Stéphane Mazouffre, Michael Keidar, and Kateryna Bazaka.
Mars colonization: Beyond getting there. Global Challenges, 3(1):1800062, 2019.
[5] Dirk Schulze-Makuch and P. Davies. Destination mars: Colonization via initial one-way
missions. Journal of the British Interplanetary Society, pages 11–14, 01 2013.
[6] NASA. Chronology of mars exploration.
[7] Cordes Florian, Kirchner Frank, and Babu Ajish. Design and field testing of a rover with an
actively articulated suspension system in a mars analogue terrain. Journal of Field Robotics,
35(7), 2018.
[8] DFKI. Rimres: Reconfigurable integrated multi robot exploration system.
https://robotik.dfki-bremen.de/en/research/projects/rimres.html, last accessed:
06/05/2019.
[9] DFKI. Sherpa: Expandable rover for planetary applications. https://robotik.
dfki-bremen.de/en/research/robot-systems/sherpa.html, last accessed: 06/05/2019.
[10] Florian Cordes, Christian Oekermann, Ajish Babu, Daniel Kuehn, Tobias Stark, and Frank
Kirchner. An active suspension system for a planetary rover. In 12th International Symposium
on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS’14), Montreal,
Canada, June 2014.
[11] DFKI. Sherpatt. https://robotik.dfki-bremen.de/en/research/robot-systems/
sherpatt.html, last accessed: 06/05/2019.
[12] DFKI. Sherpauw. https://robotik.dfki-bremen.de/en/research/robot-systems/
sherpauw.html, last accessed: 21/10/2019.
[13] Pantelis Poulakis, Jorge Vago, Damien Loizeu, Celia Vicente-Arevalo, Ryan McCoubrey, Jose
Arnedo-Rodriguez, Benjamin Boyes, Sean Jessen, Angel Otero-Rubio, Sthephen Durrant,
Garry Gould, L. Joudrier, Yuri Yushtein, Coralie Alary, Eric Zekri, P. Baglioni, Alberto
Cernusco, Fabio Maggioni, Ricardo Yague, and Franco Ravera. Overview and development
status of the exomars rover mobility subsystem. 05 2015.
[14] Haibo Gao, Zongquan Deng, Keiji Nagatani, and Kazuya Yoshida. Experimental study
and analysis on driving wheels’ performance for planetary exploration rovers moving in
deformable soil. Journal of Terramechanics, 48:27–45, 02 2011.
[15] Scott Moreland. Traction processes of wheels in loose, granular soil (phd thesis). Carnegie
Mellon University, Pittsburgh, PA, 2015.

86
[16] K. Skonieczny, S. J. Moreland, and D. S. Wettergreen. A grouser spacing equation
for determining appropriate geometry of planetary rover wheels. In 2012 IEEE/RSJ
International Conference on Intelligent Robots and Systems, pages 5065–5070, Oct 2012.
[17] S. Haggart and J. Waydo. The mobility system wheel design for nasa’s mars science
laboratory mission. In 11th European Conference of the International Society for
Terrain-Vehicle Systems, November 2008, Torino Italy, Nov 2008.
[18] JPL NASA. Mars curiosity rover raw images. https://mars.nasa.gov/msl/multimedia/
raw-images/?order=sol+desc%2Cinstrument_sort+asc%2Csample_type_sort+asc%2C+
date_taken+desc&per_page=50&page=0&mission=msl, last accessed: 21/10/2019.
[19] Nildeep Patel, Richard Slade, and Jim Clemmet. The exomars rover locomotion subsystem.
Journal of Terramechanics, 47(4):227–242, 2010.
[20] Kojiro Iizuka and Takashi Kubota. Study of flexible wheels for lunar exploration rovers:
Running performance of flexible wheels with various amount of deflection. Journal of Asian
Electric Vehicles, Volume 7(Number 2):1319–1324, December 2009.
[21] Olaf Kroemer, D. Beermann, Florian Cordes, Caroline Lange, Benjamin Littau, Roland
Rosta, Marco Scharringhausen, Tim van Zoest, and Christian Grimm. Adaptive flexible
wheels for planetary exploration. In 62nd International Astronautical Congress (IAC’11),
Cape Town, South Africa, October 2011.
[22] Brian Wilcox and Tam Nguyen. Sojourner on mars and lessons learned for future planetary
rovers. 07 1998.
[23] J. Matijevic. Sojourner - the mars path[U+FB01]nder microrover flight experiment. Jet
Propulsion Laboratory - California Institut of Technology, 1997.
[24] Mark Lapin Rick D. Smith, Robert D. Galletly. Mars pathfinder mission pioneers new
techniques, presents challenges for test engineers. NASA, JPL, 1996.
[25] Dale Ferguson, Joseph Kolecki, Mark Siebert, David Wilt, and Jacob Matijevic. Evidence for
martian electrostatic charging and abrasive wheel wear from the wheel abrasion experiment
on the pathfinder sojourner rover. Journal of Geophysical Research, 104:8747–8789, 04 1999.
[26] Jerome Johnson, Anton Kulchitsky, Paul Duvoy, Karl Iagnemma, Carmine Senatore,
Raymond Arvidson, and Jeffery Moore. Discrete element method simulations of mars
exploration rover wheel performance. Journal of Terramechanics, 23, 03 2015.
[27] NASA/JPL-Caltech/MSSS. Rover wheels. https://mars.nasa.gov/mer/mission/rover/
wheels-and-legs/, last accessed: 18/12/2019.
[28] NASA’s Mars Exploration. Mer spotlight: Wheels in the sky. https://www.nasa.gov/
vision/universe/roboticexplorers/rover_wheels.html, last accessed: 21/10/2019.
[29] S. Squyres, Raymond Arvidson, D. Bollen, J. Bell, J. Brückner, Nathalie Cabrol, W. Calvin,
Michael Carr, P. Christensen, B. Clark, L. Crumpler, David Des Marais, Claude D’Uston,
Thanasis Economou, Joseph Farmer, William Farrand, W. Folkner, R. Gellert, Timothy
Glotch, and A.S. Yen. Overview of the opportunity mars exploration rover mission to
meridiani planum: Eagle crater to purgatory ripple. Journal of Geophysical Research, 12
2006.
[30] Raymond Arvidson, J. Bell, P. Bellutta, Nathalie Cabrol, Jeffrey Catalano, J Cohen,
L. Crumpler, David Des Marais, Tara Estlin, William Farrand, R Gellert, John Grant,
Rebecca Greenberger, Edward Guinness, K. Herkenhoff, Jennifer Herman, K. Iagnemma,
J. Johnson, Goestar Klingelhoefer, and A. Yen. Spirit mars rover mission: Overview and
selected results from the northern home plate winter haven to the side of scamander crater.
Journal of Geophysical Research, 115, 09 2010.
[31] JPL NASA, Mars Exploration Programme. Mars curiosity rover. https://mars.nasa.gov/
msl/home/, last accessed: 17/12/2019.
[32] J.P.Grotzinger M.C.Heverly J.Shechet J.Moreland M.A.Newby N.Stein A.C.Steffy F.Zhou
A.M.Zastrow A.R.Vasavada A.A.Fraeman E.K.Stilly R.E.Arvidson, P.DeGrosse Jr. Relating

87
 geologic units and mobility system kinematics contributing to curiosity wheel damage at gale
crater, mars. Journal of Terramechanics, 73:73–93, 2017.
[33] Luz M. Calle. Corrosion on mars: Effect of the mars environment on spacecraft materials.
NASA/Technical Publication—2019–220238, 2019.
[34] J.R. Davis. Alloying: Understanding the Basics, chapter p351-416. ASM International, 2001.
[35] Olivier Toupet, Jeffrey Biesiadecki, Arturo Rankin, Amanda Steffy, Gareth Meirion-Griffith,
Dan Levine, Maximilian Schadegg, and Mark Maimone. Traction control design and
integration onboard the mars science laboratory curiosity rover. 03 2018.
[36] NASA/Jet Propulsion Laboratory. Lesson number 22401: Premature wear of the msl wheels.
https://llis.nasa.gov/lesson/22401, last accessed: 20/12/2019.
[37] NASA/JPL-Caltech/MSSS. Break in raised tread on curiosity wheel. https://www.nasa.
gov/jpl/msl/mars-rover-curiosity-20131220/, last accessed: 17/12/2019.
[38] NASA/JPL-Caltech/MSSS. Break in raised tread on curiosity wheel. https:
//mars.nasa.gov/resources/break-in-raised-tread-on-curiosity-wheel/, last
accessed: 17/12/2019.
[39] NASA/JPL-Caltech/MSSS. Mission updates. https://mars.nasa.gov/msl/
mission-updates/?month=8&year=2014, last accessed: 17/12/2019.
[40] NASA Glenn Research Center. Reinventing the wheel. https://www.nasa.gov/specials/
wheels/, last accessed: 21/10/2019.
[41] Damon Delap Vivake Asnani and Colin Creager. The development of wheels for the lunar
roving vehicle. NASA/TM—2009-215798, December 2009.
[42] Jason Davis. Chang’e-4 deploys rover on far side of the moon. https://www.planetary.
org/blogs/jason-davis/change4-deploys-yutu-2.html, Last accessed 31/12/2019.
[43] H˚akan Svedhem Elliot Sefton-Nash Albert Haldemann Pietro Baglioni Jorge Vago,
Daniel Rodionov and Rover Science Operations Working Group. Exomars 2020. 2019.
[44] K Farley, M Schulte, David Beaty, and Deborah Bass. Nasa’s mars 2020 rover mission. 01
2014.
[45] NASA/JPL-Caltech/MSSS. Mars 2020 mission overview. https://mars.nasa.gov/
mars2020/mission/overview/, last accessed: 18/12/2019.
[46] NASA/JPL-Caltech. Mars 2020 rover gets its wheels. https://www.jpl.nasa.gov/news/
news.php?feature=7432, last accessed: 20/12/2019.
[47] NASA/Mars 2020 Mission. Mars 2020: Wheels and legs. https://mars.nasa.gov/
mars2020/mission/rover/wheels/, last accessed: 20/12/2019.
[48] NASA/Mars 2020 Mission. Mars 2020: Body. https://mars.nasa.gov/mars2020/
mission/rover/body/, last accessed: 20/12/2019.
[49] NASA/JPL-Caltech. Mars 2020 rover is roving. https://mars.nasa.gov/resources/
24732/mars-2020-rover-is-roving/, last accessed: 26/01/2020.
[50] Dae-Young Lee, Gwang-Pil Jung, Min-Ki Sin, Sung-Hoon Ahn, and Kyu-Jin Cho.
Deformable wheel robot based on origami structure. pages 5612–5617, 05 2013.
[51] Laksh Raura, Andrew Warren, and Jekan Thangavelautham. Spherical planetary robot for
rugged terrain traversal. IEEE Aerospace Conference Proceedings, 03 2017.
[52] Yi Sun and Shugen Ma. Decoupled kinematic control of terrestrial locomotion for an
epaddle-based reconfigurable amphibious robot. pages 1223 – 1228, 06 2011.
[53] A. Najmuddin, Y. Fukuoka, and S. Aoshima. Development of rover designed to climb steep
slope of uncompacted loose sand. In 2014 IEEE/SICE International Symposium on System
Integration, pages 609–614, Dec 2014.

88
[54] J. Morales, Jorge Martínez, Anthony Mandow, Javier Serón, Alfonso Garcia, and Alejandro
Pequeño-Boter. Center of gravity estimation and control for a field mobile robot with a heavy
manipulator. IEEE 2009 International Conference on Mechatronics, ICM 2009, 04 2009.
[55] Kojiro Iizuka, Tatsuya Sasaki, Satoshi Suzuki, Takashi Kawamura, and Takashi Kubota.
Study on grouser mechanism to directly detect sinkage of wheel during traversing loose soil
for lunar exploration rovers. ROBOMECH Journal, 1:15, 10 2014.
[56] Junpeng Shao, Tianhua He, Jin-Gang Jiang, and Yongde Zhang. Recent patents on
omni-directional wheel applied on wheeled mobile robot. Recent Patents on Mechanical
Engineering, 9:215–221, 08 2016.
[57] Chris Brunskill, Nildeep Patel, Thibault Gouache, Gregory Scott, Chakravarthini Saaj,
Marcus Matthews, and Liang Cui. Characterisation of martian soil simulants for the exomars
rover testbed. Journal of Terramechanics, 46:419–438, 12 2011.
[58] M. Golombek, A. Haldemann, N. Forsberg-Taylor, Erin DiMaggio, R. Schroeder, B. Jakosky,
M. Mellon, and J. Matijevic. Rock size-frequency distributions on mars and implications for
mars exploration rover landing safety and operations. Journal of Geophysical Research, 108,
12 2003.
[59] NASA. Welcome to the planets: Valles marineris. https://pds.jpl.nasa.gov/planets/
captions/mars/marscany.htm, last accessed: 20/12/2019.
[60] J. Plescia. Morphometric properties of martian volcanoes. Journal of Geophysical Research,
109, 03 2004.
[61] Ashwin Vasavada, Allen Chen, Jeffrey Barnes, P. Burkhart, Bruce Cantor, Alicia
Dwyer-Cianciolo, Robin Fergason, David Hinson, Hilary Justh, David Kass, Stephen Lewis,
Michael Mischna, James Murphy, Scot Rafkin, Daniel Tyler, and Paul Withers. Space sci rev
assessment of environments for mars science laboratory entry, descent, and surface operations.
Space Science Reviews, 170, 09 2012.
[62] P L Read, S R Lewis, and D P Mulholland. The physics of martian weather and climate: a
review. Reports on Progress in Physics, 78(12):125901, nov 2015.
[63] P. Rogberg, Peter Read, and Stephen Lewis. Assessing atmospheric predictability on mars
using numerical weather prediction and data assimilation. Quarterly Journal of the Royal
Meteorological Society, 136, 07 2010.
[64] Huiqun Wang and Mark Richardson. The origin, evolution, and trajectory of large dust
storms on mars during mars years 24–30 (1999–2011). Icarus, 251, 05 2015.
[65] NASA/JPL/USGS (image) Emily Lakdawalla (map excluding the potential mission
locations). Map of all mars landing sites, failed and successful. https://www.planetary.
org/multimedia/space-images/charts/mars_landing_site_map_lakdawalla.html,
Last accessed 01/01/2020.
[66] Sanjoy Som, David Montgomery, and Harvey Greenberg. Scaling relations for large martian
valleys. J. Geophys. Res, 114, 02 2009.
[67] R. Jaumann, Dennis Reiss, S. Frei, F. Scholten, K. Gwinner, T. Roatsch, K.-D Matz,
E. Hauber, V. Mertens, H. Hoffmann, James Head, H. Hiesinger, Michael Carr, G. Neukum,
and HRSC Team. Martian valley networks and associated fluvial features as seen by the
mars express high resolution stereo camera. pages CD–ROM, 01 2005.
[68] NASA/JPL/Malin Space Science Systems. Nanedi vallis: Sustained water flow? https:
//www.jpl.nasa.gov/spaceimages/details.php?id=PIA01169, Last accessed 31/12/2019.
[69] Sylvain Bouley, Veronique Ansan, N. Mangold, Philippe Masson, and G. Neukum. Study of
naktong vallis, mars: Sustained fluvial activity extending during the hesperian period. 07
2007.
[70] HiRISE. East termination of naktong vallis. https://www.uahirise.org/ESP_043134_
1800, Last accessed 01/01/2020.

89
[71] Mike Mellon. Stunning landscape near mamers valles. https://hirise.lpl.arizona.edu/
ESP_028313_2220, Last accessed 31/12/2019.
[72] HiRISE. esp0 283132 220. https://hirise-pds.lpl.arizona.edu/PDS/EXTRAS/RDR/ESP/
ORB_028300_028399/ESP_028313_2220/ESP_028313_2220_RED.browse.jpg, Last accessed
01/01/2020.
[73] Jeffrey Andrews-Hanna. The formation of valles marineris: 1. tectonic architecture and
the relative roles of extension and subsidence. Journal of Geophysical Research (Planets),
117:3006–, 03 2012.
[74] Jeffrey Andrews-Hanna. The formation of valles marineris: 2. stress focusing along the buried
dichotomy boundary. Journal of Geophysical Research (Planets), 117:4009–, 04 2012.
[75] Jeffrey Andrews-Hanna. The formation of valles marineris: 3. trough formation through
super-isostasy, stress, sedimentation, and subsidence. Journal of Geophysical Research
(Planets), 117:6002–, 06 2012.
[76] J. Grant and M. Golombek. Selecting landing sites for the 2003 mars exploration rovers. 01
2004.
[77] Joseph Levy, James Head, and David Marchant. Lineated valley fill and lobate debris apron
stratigraphy in nilosyrtis mensae, mars: Evidence for phases of glacial modification of the
dichotomy boundary. Journal of Geophysical Research, 112, 08 2007.
[78] Andreas Johnsson, Jan Raack, and E. Hauber. Possible recessional moraines in the nilosyrtis
mensae region, mars. 06 2019.
[79] HiRISE. Layers in nilosyrtis mensae. https://hirise.lpl.arizona.edu/ESP_037251_
2120, Last accessed 01/01/2020.
[80] M. G. Bekker. Theory of land locomotion: the mechanics of vehicle mobility. Ann Arbor:
University of Michigan Press, 1962.
[81] Genya Ishigami, Akiko Miwa, Keiji Nagatani, and Kazuya Yoshida. Terramechanics-based for
steering maneuver of planetary exploration rovers on loose soil. J. Field Robotics, 24:233–250,
03 2007.
[82] William Clarke Smith. Modeling of wheel-soil interaction for small ground vehicles operating
on granular soil (phd thesis). The University of Michigan, 2014.
[83] Jerome Johnson, Anton Kulchitsky, Paul Duvoy, Karl Iagnemma, Carmine Senatore,
Raymond Arvidson, and Jeffery Moore. Discrete element method simulations of mars
exploration rover wheel performance. Journal of Terramechanics, 23, 03 2015.
[84] Yong-Ho Yoo, Jakob Schwendner, Sebastian Bartch, and Frank Kirchner. Real-time
capable soil contact model for simulation of legged robot on planetary surface. Journal
of Terramechanics, 2016.
[85] NASA/JPL-Caltech. Test rover aids preparations in california for curiosity rover on mars.
https://www.nasa.gov/mission_pages/msl/multimedia/pia15682.html, last accessed:
20/12/2019.
[86] M. Golombek, J. Grant, Tim Parker, D. Kass, J. Crisp, S. Squyres, M. Adler, A. Haldemann,
Michael Carr, and A. Arvidson. Mars exploration rover landing site selection. 02 2003.
[87] M. Golombek, J. Grant, Devin Kipp, A. Vasavada, R. Kirk, R. Fergason, P. Bellutta,
Fred Calef, K. Larsen, Y. Katayama, Andres Huertas, R. Beyer, Allen Chen, Tim Parker,
B. Pollard, S. Lee, Y. Sun, R. Hoover, H. Sladek, and Michael Watkins. Selection of the
mars science laboratory landing site. Space Science Reviews, 170:641–737, 09 2012.
[88] John Grant, Matthew Golombek, Sharon Wilson Purdy, Kenneth Farley, Kenneth Williford,
and Al Chen. The science process for selecting the landing site for the 2020 mars rover.
Planetary and Space Science, 07 2018.
[89] NASA / JPL / GSFC / Arizona State University. Google mars, nilosyrtis mensae.

90
 [90] U.S. Geological Survey. Topographic map of mars. https://pubs.usgs.gov/imap/i2782/
i2782_sh1.pdf, last accessed: 06/01/2020.
[91] NASA. Mars trek, nilosyrtis mensae. https://trek.nasa.gov/mars/#v=0.1&x=84.
1611964762815&y=31.68600246413978&z=8&p=urn%3Aogc%3Adef%3Acrs%3AEPSG%3A%
3A104905&d=, Last accessed 06/01/2020.
[92] Masataku Sutoh, Kenji Nagaoka, Keiji Nagatani, and Kazuya Yoshida. Design of wheels with
grousers for planetary rovers traveling over loose soil. Journal of Terramechanics, 50:345–353,
10 2013.
[93] Hiroaki Inotsume, Krzysztof Skonieczny, and David Wettergreen. Analysis of grouser
performance to develop guidelines for design for planetary rovers. 06 2014.
[94] Nildeep Patel, Richard Slade, and Jim Clemmet. The exomars rover locomotion subsystem.
Journal of Terramechanics, 47:227–242, 08 2010.
[95] Gong. Discussions on localization capabilities of msl and mer rovers. Annals of GIS, 21, 03
2015.
[96] Kojiro Iizuka and Takashi Kubota. Study on effect of grousers mounted flexible wheel for
mobile rovers. Journal of Asian Electric Vehicles, 10:1591–1597, 01 2012.
[97] Purushothaman Doss, Krishna Kaushik Yanamundra, Gokul Krishnan, and Periasamy
Chinnapalaniandi. Study of surface roughness and cutting force in machining for 6068
aluminium alloy. IOP Conference Series: Materials Science and Engineering, 346:012074,
04 2018.
[98] Sandmeyer Steel Ccompany. Specification sheet: Alloy 310/310s/310h. https://www.
sandmeyersteel.com/images/310-spec-sheet.pdf, Last accessed 21/12/2019.
[99] MatWeb. Material property data. http://www.matweb.com/, Last accessed 21/12/2019.
[100] A. Hurlich. Low temperature metals. McGraw-Hill Publishing Go., Inc., 136, 07 1963.
[101] E. David. An active suspension system for a planetary rover. In 12th International Scientific
Conference Achievements in Mechanical and Materials Engineering, Poland, June 2003.
[102] Gasparini Industries S.r.l. Metals and materials for low temperatures
and cryogenic applications. https://www.gasparini.com/en/blog/
metals-and-materials-for-low-temperatures/, last accessed: 31/01/2020.
[103] Aalco Metals Ltd. Aluminium alloy 2014a t651 sheet and plate. http://www.aalco.co.uk/
datasheets/Aluminium-Alloy-2014A-T651-Sheet-and-Plate_295.ashx, Last accessed
21/12/2019.
[104] Aalco Metals Ltd. Aluminium alloy - commercial alloy - 6061 - t6 extrusions. http:
//www.aalco.co.uk/datasheets/Aluminium-Alloy-6061-T6-Extrusions_145.ashx, last
accessed: 27/01/2020.
[105] Aircraft Materials. Aluminium alloy 7075. https://www.aircraftmaterials.com/data/
aluminium/7075.html, Last accessed 21/12/2019.
[106] Rederans Metal. Ams 4089 / alzn5.5mgcu / 7475 / t7651. http://referansmetal.com/
alasimli-aluminyum/product/75/ams-4089-alzn5-5mgcu-7475-t7651?lang=en, Last
accessed 21/12/2019.
[107] AZo Materials. Grade 6 ti-5al-2.5sn alloy (uns r54520). https://www.azom.com/article.
aspx?ArticleID=9297, Last accessed 21/12/2019.
[108] G. Critchlow, I. Ashcroft, T. Cartwright, and D. Bahrani. Anodising aluminum alloy, May 1
2005. US7922889B2.
[109] Władysław Skoneczny. Model of structure of al2o3 layer obtained via hard anodising method.
Surface Engineering, 17:389–392, 10 2001.

91
[110] A. Rajendra, Biren J. Parmar, A. K. Sharma, H. Bhojraj, M. M. Nayak, and K. Rajanna.
Hard anodisation of aluminium and its application to sensorics. Surface Engineering,
21:193–197, 2005.
[111] F. J. Humphreys and M. Hatherly. Title: Recrystallization and Related Annealing
Phenomena. 2nd ed. Elsevier, 2004.
[112] Paul Beck. Annealing of cold worked metals. Advances in Physics, 3:245–324, 06 1954.

92