Dynamic Event-Triggered Robust Optimal Tracking Control for Lower Limb Exoskeletons With Human-Robot Conflicts

This paper presents an adaptive critic neural network algorithm to address the robust optimal tracking control problem of lower-limb exoskeleton robots under non-zero equilibrium conditions. By incorporating an anomaly score into the robot's impedance model, the safety and stability of the system in human-robot interaction conflict environment are enhanced. A nominal system is introduced to transform the robust tracking control problem into an optimal tracking control problem. In order to overcome the zero equilibrium constraint of traditional Adaptive Dynamic Programming (ADP), a new cost function is designed. Considering the difficulties in solving the Hamilton-Jacobi-Bellman (HJB) equation, a critic neural network framework is employed to approximate the solution of the HJB equation. Meanwhile the controller updates driven by a dynamic event-triggered mechanism, significantly conserving computational resources. The effectiveness of the proposed algorithm is verified through simulation experiments.