Reinforcement Learning-Based Fixed-Time Trajectory Tracking Control for Uncertain Robotic Manipulators With Input...

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Robotic manipulators often face challenges with uncertainties and actuator limitations.
• Achieving precise trajectory tracking in a fixed time is crucial for many robotic applications.
• Input saturation in actuators can degrade control performance and system stability.

Purpose of the Study:

• To develop a fixed-time trajectory tracking control method for uncertain robotic manipulators.
• To address the issue of input saturation in robotic joint torque actuators.
• To leverage reinforcement learning (RL) for enhanced control strategy generation.

Main Methods:

• Implementation of a reinforcement learning (RL) control algorithm using radial basis function (RBF) neural networks (NNs).
• Utilization of a nonsingular fast terminal sliding mode technique for fixed-time error convergence.
• Design of a nonlinear antiwindup compensator to mitigate actuator saturation effects.

Main Results:

• The proposed method ensures tracking error convergence within a fixed time, with an estimated upper bound.
• The nonlinear antiwindup compensator effectively compensates for real-time actuator saturation.
• Stability analysis using a Lyapunov candidate confirms the fixed-time convergence of the closed-loop system.

Conclusions:

• The developed RL-based control law demonstrates effectiveness and superiority for uncertain robotic manipulators with input saturation.
• The combination of RL, RBF NNs, and sliding mode control offers a robust solution for fixed-time trajectory tracking.
• Simulation and experimental validation confirm the practical applicability of the proposed control strategy.