A Learning Control Strategy for Robot-assisted Bathing via Impedance Sliding Mode Technique With Non-repetitive Tasks

This paper investigates an impedance-based iterative learning sliding mode control scheme for robot-assisted bathing, taking into consideration scenarios with unknown model parameters. Initially, the utilization of impedance control is not confined to merely adjusting the desired trajectory but is also instrumental in ensuring active compliance control during the robot-assisted bathing procedure. Furthermore, an iterative learning control (ILC) is devised to estimate the iteration-invariant dynamic parameters, which are intricate and challenging to precisely ascertain in practical applications. To mitigate the effect of divergent initial conditions in ILC, a trajectory reconstruction method is introduced, thus ensuring the convergence of tracking errors even when starting from random initial states. Moreover, an adaptive sliding mode control mechanism is proposed to counteract non-parametric external disturbances and the torque generated through human-machine interaction during the bathing process. The convergence of the double closed-loop system in both the time and iterative domains is demonstrated through the application of the composite energy function method. Eventually, the efficacy and superiority of the control strategy outlined in this paper are jointly verified through co-simulation employing MATLAB and ADAMS.