Adjustable Compliance and Force Feedback as Key Elements for Stable and Efficient Hopping

Abstract

Achieving efficient and human-like hopping motions with consistent consecutive patterns requires proper mechanical design and control. In this regard, we introduce a new bioinspired design and control approach comprised of a hybrid actuation system and force-based compliance control. Combining an electric motor with a pneumatic artificial muscle makes up the so-called Electric-Pneumatic Actuation (EPA) system which enables presetting the leg compliance. Using the ground reaction force for online adjustment of EPA allows controlling the leg stiffness. This approach combines a simplified version of reflex-based control and physical impedance adjustment, that resembles the human neuromuscular system in a parsimonious way. By optimizing PAM pressures and control gains for different targeted hopping heights using Bayesian optimization, we show experimentally that the proposed approach generates various stable hopping motions and yields efficient performance. Moreover, our method is shown to be capable of creating more human-like hopping patterns compared to the previous studies on the same robot.