Impedance control of series elastic actuators based on well-defined force dynamics

Abstract

Modern rehabilitation and assistive robots are usually designed with impedance-controlled compliant actuators. Impedance control is usually implemented based on an inner force loop which assumed to be very fast. Unfortunately force control performance can be influenced by the human dynamics leading to inaccurate impedance rendering. In our previous work we solved this force control issue by proposing a human-adaptive force controller which guarantees predictable performance in spite of uncertainties in the human. In this paper we propose a robustified human-adaptive control law and ensuring asymptotic stability instead of globally uniformly ultimately boundedness. Then, we analyze the application of human-adaptive force control on impedance rendering. We show (i) that impedance accuracy is improved with respect to standard solutions and (ii) that, leveraging on human-adaptive control, impedance accuracy may not need high bandwidth inner force control. Simulation and experimental results validate the proposed method and compare it with a widely used impedance control algorithm.