Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. Automation and Machine Learning
  4. Vol 7, Issue 1, 2026

A Comparative Study of Basic Reinforcement Learning Algorithms for Two-Wheeled Mobile Robot Path Tracking

Download as PDF

DOI: 10.23977/autml.2026.070101 | Downloads: 0 | Views: 38

Author(s)

Hongyuan Liu 1

Affiliation(s)

1 School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan, China

Corresponding Author

Hongyuan Liu

ABSTRACT

To address the adaptability problem of two-wheeled mobile robots in path tracking tasks under different scenarios and improve the accuracy and robustness of robot motion control, this paper builds a path tracking simulation platform for two-wheeled mobile robots based on Gym and Gazebo. Three classic basic reinforcement learning algorithms, namely Q-Learning, Sarsa, and Deep Q-Network (DQN), are selected for comparative experimental research. Three map scenarios with varying complexity—simple, complex, and dynamic obstacle environments—are designed to conduct quantitative analysis and qualitative evaluation of the algorithms' performance from three core dimensions: convergence speed, control stability, and environmental robustness. Experimental results show that the Q-Learning algorithm converges fastest in simple static scenarios but has insufficient robustness; the Sarsa algorithm exhibits superior safe exploration capabilities; and the DQN algorithm demonstrates remarkable adaptive advantages in complex dynamic scenarios. This paper finally provides clear algorithm selection recommendations for different application scenarios, offering theoretical reference and technical support for the engineering practice of two-wheeled mobile robot path tracking control.

KEYWORDS

Reinforcement Learning; Two-Wheeled Mobile Robot; Path Tracking; Gym/Gazebo; Algorithm Comparison; Convergence Speed; Robustness

CITE THIS PAPER

Hongyuan Liu. A Comparative Study of Basic Reinforcement Learning Algorithms for Two-Wheeled Mobile Robot Path Tracking. Automation and Machine Learning (2026) Vol. 7: 1-8. DOI: http://dx.doi.org/10.23977/autml.2026.070101.

REFERENCES

[1] Mnih, Volodymyr. "Playing atari with deep reinforcement learning." arXiv preprint arXiv:1312.5602 (2013).
[2] Rummery, Gavin A., and Mahesan Niranjan. On-line Q-learning using connectionist systems. Vol. 37. Cambridge, UK: University of Cambridge, Department of Engineering, 1994.
[3] Brockman, Greg, et al. "Openai gym." arXiv preprint arXiv:1606.01540 (2016).
[4] Lillicrap, Timothy P., et al. "Continuous control with deep reinforcement learning." arXiv preprint arXiv:1509.02971 (2015).
[5] Littman, Michael L. "Markov games as a framework for multi-agent reinforcement learning." Machine learning proceedings 1994. Morgan Kaufmann, 1994. 157-163.
[6] Busoniu, Lucian, et al. Reinforcement learning and dynamic programming using function approximators. CRC press, 2017.

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 4717
Visits: 231753

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.