2021 International Seminar on Intelligent Technology and Its Applications (ISITIA)

Vacuum Cleaner Robot with Staircase Cleaning

Feature and Boustrophedon Path Planning
Gregorius Gery Gavindra Hendra Kusuma Tasripan
2021 International Seminar on Intelligent Technology and Its Applications (ISITIA) | 978-1-6654-2847-7/21/$31.00 ©2021 IEEE | DOI: 10.1109/ISITIA52817.2021.9502216

Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering

Institut Teknologi Sepuluh Nopember Institut Teknologi Sepuluh Nopember Institut Teknologi Sepuluh Nopember
Surabaya, Indonesia Surabaya, Indonesia Surabaya, Indonesia
ggerygavindra@gmail.com hendraks@ee.its.ac.id tasripan@ee.its.ac.id

Abstract—Home automation technology is rising and Numerous additional features have been developed thus far
growing in demand, one of the devices being actively such as wet mopping, auto docking, UV sterilization, house
developed and highly sought is a vacuum cleaner robot. mapping, scheduled cleaning and many more, all this besides
Products available today has many features, but none has some standard features such as anti-bump and anti-fall. These
the ability to clean on different leveled surface area or features make the vacuum robot have the ability to clean
multistory house. This research develops a vacuum single-story house flawlessly without human interference
cleaner robot that is capable of cleaning open-space area whatsoever.
using boustrophedon path planning and has the ability to Challenge arise when it is faced with a multistory house or
completely clean staircase area. Arduino Mega 2560 is when one story has different leveled surfaces. In this case, the
used as the main processor, with timing belt system to vacuum robot will not be able to operate effectively to
move a platform as the climbing mechanism. The robot completely clean all surfaces. Multiple robots or human
uses mecanum wheels, enabling the robot to move in all assistance will be needed to solve this problem, which is not
directions within a tight space. Step up test will be cost-effective and against the main purpose of home
conducted in 2 phases, first the robot will climb a box with automation.
varying height from 5-25cm with 5cm increments, The research will result in a vacuum robot that is capable
repeated 10 times for each different height. The second of climbing up a staircase and different level surfaces. The
phase tests the robot on a staircase for 20 repetitions, to robot will also completely cover the staircase area to clean
examine its ability in completely covering a staircase. The each of the steps while going up the ladder. With this
result is the robot successfully climbs to the height of additional feature, the vacuum robot will have the necessary
20cm with 100% success rate and fails at 25cm, it also ability to be implemented in either a multistory house or one
achieves 80% success rate on staircase complete area with different leveled surfaces on each story.
coverage test. The boustrophedon path-planning test is
conducted using an unobstructed, walled, 120x120cm II. BACKGROUND
open-space arena, marked with 30x30cm grid. The robot This chapter contains background information and short
is tested by placing it at 4 different initial position, explanation on theories and components used to build the
repeated 10 times for each to test its ability to find the robot.
nearest corner, and then, using the same arena, make the
robot cover it completely using boustrophedon path A. Robotic Vacuum Cleaner
planning. This results in 100% success rate in finding the Robotic vacuum cleaner is an autonomous robot that has
nearest corner and 85% success rate for complete area intelligent programming and limited cleaning capability. The
coverage. Vacuum cleaning ability is then tested by increasing number of companies that produce this device is a
having the robot clean traces of flour and salt, proof of its popularity.
representing very fine and small particles of dirt, but due
to a bad design for the dust collection tunnel, the vacuum
suction power is very weak and this results in failure to
completely clean both substances.

Keywords : Vacuum cleaner robot, staircase, boustrophedon

I. INTRODUCTION
In the era of industry 4.0, the technology of IoT devices is
rapidly growing and affects, not just industrial devices, but
also home appliances lately known as smart home devices. Fig 1. Roborock S6 and Neato Botvac D7 [2][3]
statista.com estimates by 2025, the industry of IoT device
will worth USD 75.44 billion, and up to 2018, 1.2 billion of
all IoT devices are smart home device [1]. B. Boustrophedon
One of the devices that is increasing in demand and being The word “boustrophedon” originated from ancient
actively developed is a vacuum cleaner robot. This is caused Greek meaning “like the ox turns [while plowing]” [4].
by the comfort that the device provides for busy people who
have little to no time to spare for housekeeping activities.
359

Authorized licensed use limited to: REVA UNIVERSITY. Downloaded on September 02,2025 at 13:50:40 UTC from IEEE Xplore. Restrictions apply.
 This motion is defined by the movement of going left to
right and right to left in an alternating fashion. This path-
planning algorithm is one of the simplest and most efficient
algorithms for complete area coverage.

Fig 5. Belt and pulley system diagram [8]

Fig 2. Boustrophedon path panning visualization [5]

C. Ultrasonic Distance Sensor E. Mecanum Belt

Ultrasonic distance sensor is a non-contact sensor that Mecanum wheel is an omnidirectional wheel that gives a
measures the distance from the sensor to the nearest object vehicle the ability to move in all directions. The direction in
facing the front of the sensor. This is a type of active sensor which the vehicle will move is determined by the
in which it generates pilot signal and receives signal combination of each of the wheels’ rotation, it can move
reflection as it bounces off a surface [6]. forward, backward, side-ways, on 45 degree angles, and
rotate on different axis.
The pilot signal is a 40 kHz pulse signal, which is
inaudible to the human ear. The resulting distance is
calculated from the time the signal takes from the moment it
is emitted until it is received by the sensor.

Fig 6. Mecanum wheel configuration [12]

Fig 3. Ultrasonic distance sensor working principle [6] III. HARDWARE DESIGN
D. Belt Driven System This chapter explains the design of the robot, from
Belt system is a system where the transfer of energy in a mechanical system used and also the algorithm planted on the
system uses a belt and a pulley as its media. There are several robot’s microcontroller.
shapes of belt available to use, which are round, flat, v- A. Hardware Design
grooved, and synchronous belt.

Fig 4. Various shapes of timing belt [7]

The system consist of 3 main components, a source of

motion, a belt, and a pulley. The basic principle of this system
is the transfer of rotating motion from the source of to the
pulley via a timing belt. This simple mechanical system is
used in various applications due to its simplicity and ease of Fig 7. Overall system block diagram
maintenance. The overall system consists of an Arduino Mega as its
main processor with inputs from six micro switches and five
distance sensors. The system’s actuators are six DC motors
and a blower fan. The distance sensors’ placements are one
on the left and right side, two facing forward, and one on the
front platform. The micro switches are used as an anti-bump
sensor to stop the robot’s movement after it bumps into an

360

Authorized licensed use limited to: REVA UNIVERSITY. Downloaded on September 02,2025 at 13:50:40 UTC from IEEE Xplore. Restrictions apply.
object. The switches are placed one on each side directly After arriving at the nearest corner of the arena, the robot
above the distance sensors, one on the front platform, and two will start to do the boustrophedon path area coverage process.
as a stop switch on the front and back moving platform. The The arena used on this research is an unobstructed arena
blower fan will provide vacuum suction power by utilizing surrounded by 30 Cm wall. By using boustrophedon path
the vacuum created on the fan’s intake side when the fan is planning, robot will cover 100% of the area, visiting every
running. grid at least once.

Fig 8. 3D design render of the robot

The letter “D” shaped design is chosen as this design

maximizes the space inside the robot and has a flat front side.
This design has better front sensor placement compared with
a more common round design as the sensor has greater Fig 10. Boustrophedon path planning flowchart
distance in between, this helps the aligner function that
parallels the robot to the wall the robot is facing to. The flat While executing this sub-process, if the robot encounters an
front side also makes for a better climbing up mechanism as obstacle higher than 20 Cm, it will register this as a wall and
the front platform has greater width, which stabilizes the will continue the sub-process, if it is below 20cm, this will be
robot when it is climbing up an obstacle. registered as a climbable obstacle, either a different leveled
surface or the first step of a staircase. Either way, the robot
The system is specifically designed for climbing up an will then move to execute the stair-climbing sub-process.
obstacle, not climbing down. This is due to the increase in
sensors needed to perform a climbing down motion.
B. Software Design
The program is built on Arduino IDE using Arduino’s C++
programming language. It is broken down into sub-processes,
one for each of the movements the robot will use, such as
moving forward, backward, side-ways, 90 degree clockwise
and counter clockwise turn, etc. These sub-processes will be
called upon when it is needed.
The main routine starts with position initialization sub-
process. This sub-process’ goal is to find the nearest corner
of an open space area, this is done assuming that at the
beginning of the test, the robot is placed randomly within an
open space arena.

Fig 10. Boustrophedon path planning flowchart

IV. ANALYSIS
This section explains the method used to test each of the
robots function and also gives analysis based on the data
included.
A. Unobstructed Space Coverage Test
This scenario will test the device’s ability to cover an
unobstructed flat surface using boustrophedon path planning.
The test will use a 120x120 Cm arena marked with a 30x30
Cm grid surrounded by 30 Cm wall all around. Rows marked
with numbers and columns with alphabets.
Fig 9. Position initialization flowchart

361

Authorized licensed use limited to: REVA UNIVERSITY. Downloaded on September 02,2025 at 13:50:40 UTC from IEEE Xplore. Restrictions apply.
 The test will start by testing the accuracy of the corner B. Obstacle Climbing Test
finding sub-process. The robot will start at a certain grid The scenario of this test is to put a variable-height obstacle
in front of the robot and have the robot climb it. The obstacle
ranges from 5-25 Cm with 5 Cm increament.
Height (cm) Success rate
(10 repitition)
5 100%
10 100%
15 100%
20 100%
25 0%
Table III Various height climbing test result

This results in a 100% success rate in climbing an obstacle

up to the height of 20 Cm, at 25 Cm the robot fails to climb
because of mechanical limitation, but because the average
staircase step height is 19.7 Cm [13] this result is already
within the arena and execute the position initialization sub- satisfying.
process to find the nearest corner. The result is noted and The next scenario is to have the robot cover a stair arena.
compared with the expected result.
The model used to test this scenario is a staircase model with
Fig 11. Unobstructed open area test arena 3 steps with the height of 17 Cm. The step width is 40 Cm
and the stair width is 80 Cm. The test will be repeated 20
Target end Result (from 10 times for this scenario.
Initial position
position repetition)
C3 D4 D4
D3 D4 D4
A2 A1 A1
B2 A1 A1
Table I Area coverage result data

The result is combined into one table because the robot

achieved 100% success rate in finding the correct corner
throughout all 10 repetitions.
The test then continues by having the robot execute the
boustrophedon path planning sub-process. The robot is
placed on two different initial grids with 2 different
orientations on each grid. The result will be noted down as a
grid sequence in which the robot makes a 90-degree turn.
This test is repeated 10 times in each different initial Fig 12. Staircase test arena
positions.
The result is an 80% success rate for complete staircase
Initial grid Success rate area coverage. Failures happen due to the same reason as the
Grid sequence
(facing towards) (10 repetition) path-planning test.
A4, B4, B1, C1,
A1 ( A4 ) 90% C. Vacuum Suction Test
C4, D4, D1
D1, D2, A2, A3, The scenario for this test is to have the robot clean 2 kinds
A1 ( D1 ) 100% of substances with different particle sizes and examine the
D3, D4, A4
D4, C4, C1, B1, ability of the robot in cleaning those substances. The
D1 ( D4 ) 80% substances are flour (average size 200 µm) and salt (average
B4, A4, A1
A1, A2, D2, D3, size 0.3 mm), representing dust particles with very fine and
D1 ( A1 ) 70% small particle size. The test is repeated 10 times for each
A3, A4, D4
substance.
Average 85% On all the trials done, the robot fails to clean any of the
Table II Area coverage test result data substances. This is due to mechanical design failure, where
the tunnel from fan intake to dust input hole is too big, so
The test results in an 85% success rate for complete area according to Venturi effect, the pressure on the input hole is
coverage. Failure happens because of the uneven surface of
the arena, in some runs, this causes the robot to fail on
executing a 90-degree turn, resulting in failure to make the
expected sequence.

362

Authorized licensed use limited to: REVA UNIVERSITY. Downloaded on September 02,2025 at 13:50:40 UTC from IEEE Xplore. Restrictions apply.
excessively low to create the necessary vacuum needed to REFERENCES
suck any particle.
[1] Statista, "Forecast market size of the global smart home market
from 2016 to 2022", Available:
https://www.statista.com/statistics/682204/global-smart-home-
market-size/ [Accessed : 25-January-2021].
[2] World Design Guide, "Roborock S6 / Cleaning Robot", Available :
https://ifworlddesignguide.com/entry/257124-roborock-s6
[Accessed : 25-January-2021].
[3] PCMag, "Neato Botvac D7 Connected", Available :
https://uk.pcmag.com/vacuums/90981/neato-botvac-d7-connected
[Accessed : 25-January-2021].
[4] Wikipedia, "Boustrophedon",
Available :https://en.wikipedia.org/wiki/Boustrophedon [Accessed :
25-January-2021].
[5] Ntawumenyikizaba, A., Viet, H., & Chung, T. (2012). An online
complete coverage algorithm for cleaning robots based on
boustrophedon motions and A* search. 2012 8th International
Conference on Information Science and Digital Content
Technology (ICIDT2012), 2, 401-405.
[6] Fraden, Jacob, "Handbook of Modern Sensors : Physics, Designs,
Fig 13. Vacuum suction test and Applications", Advanced Monitors Corporation, USA, Ch. 7.6,
2004.
V. CONCLUSION [7] Perpetual Industries, "Types of Belt Drives", Available :
1. The robot has a 100% success rate on climbing an obstacle https://xyobalancer.com/types-of-belt-drives/ [Accessed : 25-
January-2021].
up to 20 Cm of height.
2. Area coverage test has 100% success rate on finding the [8] Robotpark Academy, "Robotic Mechanisms - Pulleys adn Belts
51045", Available : http://www.robotpark.com/academy/robotic-
nearest corner on 10 repetitions and 85% success rate for mechanisms-pulleys-and-belts-51045/ [Accessed : 25-January-
boustrophedon path planning for complete area coverage 2021].
on 10 repetitions. [9] Components101, "L298N Motor Driver Module". Available :
3. The robot has a success rate of 80% on 20 repetitions for https://components101.com/modules/l293n-motor-driver-module
complete staircase area coverage. This result is achieved [Accessed : 25-January-2021].
on the explained staircase testing rig consisting of only 3 [10] Amazon, “DC Worm Gear Motor 12V High Torque Reduction Gear
steps. Box with Encoder Srong Self-locking 6mm Output Shaft(20RPM)”,
Available : https://www.amazon.com/Torque-Reduction-Encoder-
Self-locking-Output/dp/B073S65J25 [Accessed :25-February-
Plans are being developed for future development of this 2021].
robot, with main focus on better vacuum suction tunnel [11] ELPROCUS, "Arduino Mega 2560 Board", Available :
design to provide suction power capable of cleaning fine dust https://www.elprocus.com/arduino-mega-2560-board/ [Accessed :
particle. Adding extra width to the front moving platform will 25-January-2021].
also increase robot’s stability during climbing up motion, and [12] Gobilda, "3606 Series Mecanum Wheel Set (Bearing Supported
also trial can be done for a 2-wheel design vacuum robot for Rollers, 100mm Diameter)", Available :
reducing overall weight and increase efficiency. https://www.gobilda.com/3606-series-mecanum-wheel-set-
bearing-supported-rollers-100mm-diameter/ [Accessed : 21-
February-2021].
ACKNOWLEDGEMENT [13] House Plans Helper, "Staircase Dimensions", Available :
https://www.houseplanshelper.com/staircase-dimensions.html
Authors would like to acknowledge supports provided by [Accessed : 27-January-2021].
Electrical Engineering Department ITS on research
facilities and financial support during completion of this
research.

363

Authorized licensed use limited to: REVA UNIVERSITY. Downloaded on September 02,2025 at 13:50:40 UTC from IEEE Xplore. Restrictions apply.