Robust Trajectory Tracking of Two-Link Rehabilitation Robot Under Sudden External Disturbances By GA-Optimized PID Control

Abstract

As populations age and the need for rehabilitation grows, rehabilitation robots have drawn increasing attention. These devices guide impaired limbs through precise, repeatable motions to accelerate recovery. A central difficulty, however, is guaranteeing smooth human-robot interaction when unexpected disturbances arise during exercise. This study proposes a Proportional-Integral-Derivative (PID) controller whose six gains are globally optimized by a genetic algorithm to secure robust joint tracking for a planar two-link rehabilitation robot working with patients. PID reproducible disturbance-injection interface uses the Jacobian to convert end-effector forces into joint torques, simulating sudden patient-generated perturbations. The control law adds gravity feed-forward, a filtered PID term, and explicit actuator saturation limits. The genetic algorithms (GA) fitness function mixes time-weighted tracking error, overshoot penalty, recovery time, torque rate, and saturation duration to favor robustness. MATLAB simulations show the arm returning to within ±2° in 1.5 s, exhibiting 20 % overshoot while saturation remains low; the tuned controller outperforms a manually adjusted PID. The scheme is straightforward to deploy and meets the dual demand for safety and accurate trajectory tracking in clinical use.