Unified GRF-based control for adjusting hopping frequency with various robot configurations

Abstract

The highly dynamic hybrid nature of legged locomotion makes it a very challenging task to control. A proper control strategy, besides the ability to generate stable motions, should also possess generalization capabilities and adjustability to different conditions. In this regard, this work takes a step forward in promoting the concept of Force Modulated Compliance (FMC) control as a bioinspired reflex-based control approach and assesses its potential in generating various hopping motions. The FMC is an easy-to-tune and simplified representation of the neuromuscular control, which functions as an adjustable spring modulated by the ground reaction force. We implemented the FMC controller on EPA-Hopper-II which is an extension of our previous electric-pneumatic actuation (EPA) designs. By optimizing control parameters for hopping frequencies of 1.5−3.5Hz and different PAM configurations in the robot structure, we show that FMC can generate stable hopping motions and adapt to different configurations. The optimization results also show that leg stiffness has a high correlation with normalized energy consumption. Increases in leg stiffness decrease the normalized energy consumption and vice versa. The outcomes of this study, particularly the highlighted generalization of the FMC control, open new doors for efficient control of legged robots and assistive devices.