An Adaptive LADRC Trajectory Tracking Method Based on Q-Learning for Mobile Robots

Conventional LADRC exhibits strong dis-turbance rejection capabilities and low model de-pendency. However, its performance heavily relies on manual parameter tuning. In complex environments, fixed-parameter LADRC often fails to meet the strin-gent trajectory tracking requirements of high-precision robotic systems. To overcome this limitation, this pa-per proposes an adaptive LADRC trajectory tracking algorithm based on Q-Learning. First, a Q-LADRC controller is designed based on the trajectory model of a mobile robot. Then, a Q-learning-based adaptive mechanism is introduced to enable online dynamic adjustment of LADRC parameters. Additionally, an incremental active learning exploration strategy is employed to enhance the agent's learning efficiency and adaptability. Finally, simulation results validate the superior performance of the proposed method.