Trajectory Tracking of Autonomous Excavator Based on RBF–PID–Admittance Hybrid Control
To enhance the tracking accuracy and compliance of unmanned excavators, a force–position heterogeneous dual-loop nested cooperative control strategy is proposed. The outer loop employs admittance control to establish a dynamic mapping between contact forces and position correction commands, converting abrupt force variations into compliant position compensation. The inner loop utilizes an RBF neural network for online gain scheduling of PID parameters, thereby suppressing residual tracking errors introduced by the outer loop. Furthermore, asymptotic stability of the heterogeneous dual-loop nested closed-loop system is demonstrated via the Lyapunov method. Simulation results indicate that, in terms of position control, the RMSE and MAE for all three joints remain within 1°, with steady-state tracking errors confined to ±2°; compared with conventional PID and RBF-PID controllers, the tracking accuracy is improved. Regarding force control, compliant regulation of excavation contact forces is achieved, thereby enhancing both the precision and robustness of trajectory tracking for unmanned excavators.