Ackermann mobile robot chassis with

independent rear wheel drives

J. Hrbek, T. Ripel and J. Krejsa
Brno University of Technology, Faculty of Mechanical Engineering, Brno, Czech Republic
e-mail: jan@hrbacek.info, tripel@centrum.cz, krejsa@fme.vutbr.cz

Abstract This paper describes four-wheeled robotic

chassis of the robot Bender II utilizing Ackermann steering
and independent rear drive units. Hardware and software
implementation details as well as practical experiences of
this approach deployment are presented.
Keywords Electrical Drive, Mechatronics, Motion
control, Robotics

While the foregoing chassis types are advantageous for

use in mobile robotics, the real-world vehicles (except
some working and military machines) make usually use of
a four-wheeled chassis with the Ackermann steering. It
offers the most efficient operation and good behavior at
high speeds, but also relative low maneuverability sequent
to its nonholonomic constraints.
II. BENDER II PLATFOM DESCRIPTION

I. INTRODUCTION
Design process of a mobile robot includes many
concept decisions to be made in order to choose a proper
solution of each construction aspect keeping in mind often
contradictory requirements. One of the most important is
the selection of the chassis type.
There are many different chassis concepts usable for
a mobile robot [1]. The most common is probably the
differential drive system [2] for its simple utilization and
high maneuverability, although it does not excel in
efficiency and can damage less resistant surfaces while
turning. Another type, the omnidirectional chassis [3],
provides excellent maneuverability, but requires a flat
solid surface to ensure good behavior and prevent
uncontrollable motion.

Fig. 1. Mobile robot Bender II 3D model

Bender II is a mid-sized wheeled mobile robot designed

at the Institute of Solid Mechanics, Mechatronics and
Biomechanics as a platform for autonomous delivery and
related algorithms development. It is 60 cm long, 30 cm
wide and weighs about 25 kg.
Utilization of the Ackermann steering together with
independent rear drives in Bender II is a result of an effort
to build a system conceptually similar to real vehicles (as
for example the path planning algorithms should be
designed and tested on a robot with at least comparable
maneuverability) but not suffering from the loss of
traction in case of one wheel slippage (common for
vehicles fitted with the mechanical differential). In the
further text, our modification of a classical automotive
chassis is described.
A. The swinging rear axle
In order to simplify the overall construction, several
changes to the classical automotive chassis ware made.
The largest one applies to the wheel suspension as the
common independent wheel suspension is a complex
mechanical structure and is in piece quantities hard and
expensive to manufacture, it has been decided to use
a swinging rear axle. This solution allows the chassis to
deal with terrain irregularities (through the chassis torsion)
and stay mechanically simple at the same time.
Realization of the swinging rear axle is quite
straightforward. The body of the robot is divided into two
parts, the rear (holding both drive units) and the front
(carrying the rest of the equipment). These two parts are
connected by a shaft housed in bearings that enables the
rear axle to swing. This method is usually used when
heavy loads make the classic independent suspension
impossible. The Bender II platform is also designed to
carry heavy loads (such as a laser scanner and other
sensory equipment). A torsional spring is placed between
the front and the rear part to avoid beats and for better
stability. Figure 2 shows, how the swinging axle works.
The rear part of the robot can rotate relatively to the front
part.
The rear axle has no mechanical differential fitted the
drive units of both rear wheels are independently

controlled allowing the master control system to set

different velocities for each wheel. In dependency on the
actual steering angle and the desired forward speed the
wheel velocities are computed and set to the drive units.
The particular algorithm is described below.
Each drive unit consists of the Maxon RE40 DC motor
with a 43:1 planetary gearhead and a chain transmission to
the wheel (1.5:1 ratio). The motor is controlled by
a specially designed speed controller communicating with
the master computer using a RS-485 bus. This two drive
units provide total power of 300 W and total constant
torque of 30 Nm (wheels have 15 cm in diameter). The
power is supplied by two 12 V/7 Ah sealed lead-acid
accumulators.
B. Ackermann streering
Ackermann steering (also known as kingpin steering)
solves the difference of angles between steering wheels
during vehicle turning. This difference is caused by the
fact that each wheel follows a different radius, so that the
inside wheel has to be tilted a little more than the outside
wheel. This principle radically reduces tire slippage (that
is important especially at higher speeds). It is realized by
double pivoting system, where the pivots are at an angle.
The angle has to assure that the kingpin axis, end of the
pivot and center of the rear axle are in line (as shown on
Fig. 3). It is possible to detent the static toe-in by
a threaded rod in order to achieve better driving behavior.
According to these diagrams, the kingpin is placed in
the center of the wheel; that can be very difficult or even
impossible to achieve (from the construction point of
view), mainly when common commercially available offshelf wheels are used. The Bender II front wheels center
of rotation is placed approximately 50 mm off the wheel
center.

III. SOFTWARE DIFFERENTIAL

As already discussed in the previous chapter, angular
speeds of the rear wheels (or motors) are controlled
independently by the software differential. Individual
motor angular speeds must be described in dependence on
the steering angle . It is needed to introduce angular
speed of a virtual motor fitted in the center of the robot:
v
= i
(1)
r
where v denotes forward speed, r is wheel radius and i
the total gear ratio between the motor and the wheel. From
the Figure 3 it is possible to derive the relation between
the steering angle and the curve radius R:
l
R=
(2)
tan
where l is the wheel base of the chassis. The angular
speeds of the motors are then dependent on the ratio
between the wheel spacing d and the current curve radius
R:

L = 1 +

2R

(3)

(4)

2R
where L (R) is the left (right) motor angular speed.

R = 1

On-board computer
Mission control
H
I
G
H

Environment model
(Indoor/Outdoor)

Global planner

L
E
V
E
L

Sensor data
processing

Compass
Motion

M
I
D
D
L
E

Odometry

Broadcast

Compass

Servo

Drive (left)

Drive (right)

DoubleEncoder

L
E
V
E
L

LMS

WebCam

Fig. 2. Swinging rear axle of the Bender II

SharpGPS

ImageProcessing

PathRunner

L
O
W
L
E
V
E
L

RS-485 bus

Fig. 4. Bender II software architecture

Fig. 3. Ackermann steering principle

RS-485
bus driver

Electronic
compass

Steering
servo

Front IRCs

Right drive

Left drive

SICK LMS

Camera

GPS

Serial port wrapper with queued interface

H
A
R
D
W
A
R
E

All quantities used to describe any parameter of the

robots turning in this paper are introduced following
a simple convention turning to the right implicates
positive sign of the value, turning to the left is represented
by negative sign. A similar rule is applied to velocities
positive speed means forward movement, negative implies
reversing.
A. Implementation of the software differential
As the software architecture scheme (Fig. 4) shows,
there is a module Motion that lies on the top of the lowlevel modules (individual hardware devices interfaces).
This module is an entry point for the higher software
layers to control the motion of the robot. It encapsulates
the software differential and provides two public methods
method Go(speed, direction) used to drive the
robot at the desired speed to the desired direction and
method Stop() to halt the robot.
The internal structure of the SW differential code
follows the equations presented previously. The algorithm
firstly converts the desired robot speed to the angular
speed of a virtual centered motor. Then the desired curve
radius according to (2) for a non-zero direction angle is
computed (zero has the meaning of a straight movement
and matches an infinite curve radius).
The next step is already to calculate the individual
motor angular speeds according to (3) and (4). This can be
done only when a non-infinite value of the curve radius is
provided. Otherwise the differential algorithm is skipped
and both wheels are driven at equal angular speeds.
The Motion module has now all information to order
the steering servomotor and the drive units to set the
currently computed values. This procedure repeats every
time the upper software layers decide to change desired
speed or direction of the movement.
IV. PRACTICAL EXPERIENCES
The robot Bender II has been thoroughly tested during
both indoor and outdoor localization/navigation
algorithms development using the described architecture.
It has been found that the chassis of our robot behaves
well under various conditions. The independent rear wheel
drives system does not suffer from the complete loss of
traction in case of one wheel slippage, as both drive units
hold their preset speeds independently. The robot is thus
able to drive through surprisingly hard terrains and still
behave well and economically on the road.
While the ratio of the rear wheel angular speeds is
dependent only on the steering angle and not on the
unstable adhesion of individual wheels, the robot is not
prone to under- or oversteering the Ackermann steering
is supported by the rear wheel speeds ratio.
The only disadvantage of the presented independent
drive units architecture (apart from obvious problems
arising from using two motors, two gears and two
controllers instead of one) is that the chassis becomes less
controllable in case of misbehavior of one of the drive
units. This happened several times at the beginning of the
chassis testing and was caused by communication
problems. Such problems can be avoided by using a single
controller driving both traction motors, when a possible
communication failure would not cause discrepancy in
wheel angular speeds.

Fig. 5. Bender II chassis with a testing DC motor fitted

V. CONCLUSIONS
This paper describes the use of the Ackermann steering
together with independent rear drive units controlled by a
software differential algorithm. Utilization of this solution
was selected in order to provide an energetically effective
yet robust mobile robotic platform.
Our approach has been proven to be functional under
various conditions during operation in both indoor and
outdoor environments. The use of independently driven
rear wheels with the automotive chassis is commendable
when other type of chassis is unemployable and desired
environment would bring problems with the mechanical
differential.
ACKNOWLEDGMENT
Published results were acquired with the support of the
Ministry of Education, Youth and Sports of the Czech
Republic, research plan MSM 0021630518 Simulation
modeling of mechatronic systems.
REFERENCES
[1]
[2]

[3]

R. Siegwart, I. Nourbakhsh, Introduction to Autonomous Mobile

Robots, MIT Press, 2004
G. Campion, G. Bastin, B. dAndrea-Novel, Structural properties
and classification of kinematic and dynamic models of wheeled
mobile robots, IEEE Trans. Robot. Autom. 12, pp.47-62, 1996.
J. Salih, M. Rizon, S. Yaacob, A. Adom, M. Mamat, Designing
Omni-Directional Mobile Robot with Mecanum Wheel",
American Journal of Applied Sciences 3 (5), pp.1831-1835, 2006.