Force-controller Design Based on Adaptively Switched Iterative Learning for Cable-driven Ankle Exoskeleton
The human-exoskeleton system is a strongly coupled system,and the incidental variation in gait will cause significant disturbance to the controller,resulting in oscillations within the system.To ad-dress this issue,an adaptive switching iterative learning control strategy was proposed.When the peak er-ror value was large,a peak iterative learning control strategy was employed,which only iteratively updated the peak point of the force control signal.When the peak error is small,the control strategy switched to the conventional PD-type iterative learning control,which iteratively updated all points of the force control signal.To further improve the control effectiveness,a feedforward controller was designed to compensate for the nonlinear friction of the Bowden cable transmission.Simulation and experimental results on the ex-oskeleton platform show that our peak tracking iterative learning control strategy,compared to traditional PD-type algorithms,ensures smoother assistive force curve transitions.At the same time,this strategy has a faster convergence speed and a faster recovery speed after disturbances,achieving accurate tracking control of assistive force during walking process.
万方数据
维普
