Control of a differential drive mobile robot
Location
Havener Center, Meramec Gasconade Room, 1:30pm-3:30pm
Start Date
4-1-2026 2:00 PM
End Date
4-1-2026 2:30 PM
Presentation Date
April 1, 2026; 2:00pm-2:30pm
Description
This work investigates learning-based path tracking for the Quanser QBot Platform, a differential-drive mobile robot performed in the Control Systems and Networking Laboratory at Missouri S&T. The first step of the project was to develop a model of the robot and establish a baseline control framework in MATLAB Simulink for simulation and hardware testing. Building on that foundation, a neural-network-based actor-critic controller was then implemented to improve tracking accuracy under model uncertainty. Controller performance is evaluated in both simulation and hardware using a trifolium reference trajectory chosen to introduce repeated curvature reversals and continuous heading transitions. Results show that learning-enabled configurations reduce cumulative tracking cost and improve agreement between the desired and measured robot trajectories relative to a fixed baseline. Hardware experiments also demonstrate progressive improvement over repeated trials, with the largest gains occurring between the first and fifth learning traversals.
Biography
Landon Meyer is an undergraduate senior in Electrical Engineering at Missouri S&T. He is involved in Honors program, Tau Beta Pi, Eta Kappa Nu, Kappa Alpha Order, and research under Dr. Jagannathan in the Control Systems and Networking Laboratory, where he studies modeling, control, and learning-based path tracking for the Quanser QBot Platform robot. His work focuses on improving the accuracy of mobile-robot trajectory tracking through neural-network-based control methods. Through this research, he has gained experience in MATLAB/Simulink, controller design, simulation, and hardware testing using motion-capture feedback.
Meeting Name
2026 - Miners Solving for Tomorrow Research Conference
Department(s)
Electrical and Computer Engineering
Second Department
Computer Science
Document Type
Presentation
Document Version
Final Version
File Type
text
Language(s)
English
Rights
© 2026 The Authors, All rights reserved
Control of a differential drive mobile robot
Havener Center, Meramec Gasconade Room, 1:30pm-3:30pm
This work investigates learning-based path tracking for the Quanser QBot Platform, a differential-drive mobile robot performed in the Control Systems and Networking Laboratory at Missouri S&T. The first step of the project was to develop a model of the robot and establish a baseline control framework in MATLAB Simulink for simulation and hardware testing. Building on that foundation, a neural-network-based actor-critic controller was then implemented to improve tracking accuracy under model uncertainty. Controller performance is evaluated in both simulation and hardware using a trifolium reference trajectory chosen to introduce repeated curvature reversals and continuous heading transitions. Results show that learning-enabled configurations reduce cumulative tracking cost and improve agreement between the desired and measured robot trajectories relative to a fixed baseline. Hardware experiments also demonstrate progressive improvement over repeated trials, with the largest gains occurring between the first and fifth learning traversals.
Comments
Advisor: Jagannathan Sarangapani, sarangap@mst.edu