Page 1 of 16
European Journal of Applied Sciences – Vol. 11, No. 1
Publication Date: January 25, 2023
DOI:10.14738/aivp.111.13779.
Kohlbrecher, B., & Antoun, S. (2023). Experimental Build of a Two-Wheeled Inverted Pendulum Mobile Robot. European Journal of
Applied Sciences, 11(1). 169-184.
.
Services for Science and Education – United Kingdom
Experimental Build of a Two-Wheeled Inverted Pendulum
Mobile Robot
Blake Kohlbrecher
Colorado Mesa University
Sherine Antoun
Colorado Mesa University
ABSTRACT
This undergraduate research project describes the construction, autonomous
control, and exploration of potential uses of a Two-Wheeled Inverted Pendulum
(TWIP) in uncontrolled environments [1]. Initially, the non-functioning physical
design is two repurposed wheelchair wheels and motors mounted to a chassis that
holds two batteries to power the robot. The goal was to continue with the legacy
design, augment the build to the point where the robot is programmable, and
potentially add sensors for further applications. An- other goal was to learn how to
design a safe circuit to power the robot and its controllers. Another objective was to
learn about the fundamentals of programming an autonomous robot. This paper
describes the experimental research progress and findings. The robot is curently
programmable with an Arduino Mega 2560, which controls the wheels with two
separate motor controllers. It has a limited footprint and has the potential to carry
substantial payloads, which promises to be beneficial in uncontrolled workspaces.
The initial research has the robot capable of being programmed to go in a straight
line and currently has an MPU 6050 accelerometer and gyro attached to the
Arduino, delivers state data used to make accurate turns and sense whether the
robot has accelerated too quickly and may be in danger of becoming unbalanced.
INTRODUCTION
We carried out this research in three different experimental stages. Stage one investigated
what components were viable on the legacy robot when initially received. We decided to
preserve the structural design of the robot, and examining the state of the chassis and
motors was pivotal in making sure that would still be possible. We tested the power
supply batteries and circuits this revealed the circuitry and power supply AGM batteries
were beyond salvage. We utilized new batteries to understand the wheelchair motors’
operational parameters. Wheelchair motors have built-in safety mechanisms and circuitry.
We unmounted motors and bench-tested them individually for safety purposes.
Stage two involved designing and implementing a new circuit on the TWIP that would
include a programmable controller. We accomplished this by testing with motor
controllers and an Arduino Mega 2560, other architectures were examined but this board
was selected for its simplicity, flexibility and robustness [2]. Testing at this stage with the
robot almost fully assembled wheels removed to ensure the robot had no potential
mobility during this time. While elevated on a static stand, attached wheels would still
have the potential to perturb the robot. Motion control library functions were designed
Page 2 of 16
170
European Journal of Applied Sciences (EJAS) Vol. 11, Issue 1, January-2023
Services for Science and Education – United Kingdom
and coded [3][4], actuators’ speed ramping control function provided conclusive evidence
that the robot controller could be programmed. Multiple circuit setups were employed
during this time, as different iterations revealed further requirements to operate and
control the robot or uncovered faulty circuit designs.
Figure 1. View of the motor, note manual safety break arm, front section light casing
contains servo actuated electronic braking system.
Stage three tests were on the fully assembled mobile robot and mainly focused on tuning
the functions controlling the robot. Experiments sought to quantify the difference between
the power output to the wheels and finding safe ramping intervals for accelerating and
decelerating the robot. Stage three also sought to test an emergency power shut-off switch
while moving to immediately stop the robot in case of unexpected departure from safe
operation mode due to unforeseen events. We implemented, tested, and refined a pro- grammed robot brakes application allowing the controller to make quick stops.
METHODOLOGY
Physical Design
The physical design encompasses the following list. Figures 6-9 show pictures that show
all of these components:
• Two 12 V batteries, with one in the front, and one in the back both 20 AH AGM
Duracell absorbed glass mat (AGM) batteries (Figures 3, 4).
• The robot has two pairs of 12Volts lights, one with green labels and the other
with red and are currently not powered; subsequent robot builds iterations will
indicate the robot translation direction, with green labelled lights to denote forward
translation and the red reverse (Figures 3, 4).
• Two PM802 D11F motors. These are permanent magnet DC motors of 24 Volts.
They have two pairs of power input wires. The first pair are powered to drive
the motors. The second pair is the electronically actuated brake wires. These are
connected to internal brakes, sending 24V disengaging the brakes. Experimentation
Page 3 of 16
171
Kohlbrecher, B., & Antoun, S. (2023). Experimental Build of a Two-Wheeled Inverted Pendulum Mobile Robot. European Journal of Applied Sciences,
11(1). 169-184.
URL: http://dx.doi.org/10.14738/aivp.111.13779
revealed that if the brake wires are powered, but if the main power wires to the
motor are sent 0Volts from the motor controllers,the brakes will re-engage. This
capability is used for programming motor break commands to stop the robot.
Figure 2. Faulty circuit detailing the connections between the motors, motor controllers,
batteries, and Arduino Mega 2560 that resulted in a short circuit.
Figure 3. View from the front of the robot, note battery mounting, and motor controller
interface bridges.