Reinforcement Learning-Based Tracking Control for a Three Mecanum Wheeled Mobile Robot.

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Maneuvering with Mecanum wheels presents unique challenges due to their omnidirectional capabilities.
• External disturbances can significantly impact the trajectory tracking accuracy of mobile robots.
• Robust control strategies are essential for reliable operation in uncertain environments.

Purpose of the Study:

• To develop a robust optimal tracking control strategy for a three Mecanum wheeled mobile robot (MWMR).
• To address the challenges posed by external disturbances and potential wheel slipping.
• To utilize an online actor-critic synchronous learning algorithm for adaptive control.

Main Methods:

• Establishing the Euler-Lagrange motion equation for the MWMR, considering slipping.
• Transforming the tracking control problem into an optimal control problem of an augmented system.
• Applying an online actor-critic synchronous learning algorithm to derive an approximate optimal tracking controller.
• Employing Lyapunov-based analysis to guarantee ultimately bounded tracking performance.

Main Results:

• The developed controller effectively manages external disturbances.
• The online actor-critic synchronous learning algorithm demonstrated its capability in achieving approximate optimal tracking.
• Simulation results validated the robustness and effectiveness of the proposed control strategy.
• Lyapunov analysis confirmed the guaranteed bounded tracking performance.

Conclusions:

• The online actor-critic synchronous learning algorithm provides an effective approach for robust optimal tracking control in MWMRs.
• The proposed control strategy ensures reliable trajectory following despite external disturbances and system uncertainties.
• This research contributes to the advancement of autonomous navigation systems for complex mobile robots.