Athlete Robot with applied human muscle activation patterns for bipedal running

Abstract

The essential component of legged locomotion is control of the ground reaction force. To understand the role of the musculoskeletal body in dynamic locomotion, we investigate bipedal running using a musculoskeletal “Athlete Robot”. The configuration of the muscles in the robot is compatible with the human. The spring-like property of the human lower leg during running is modeled as an elastic blade foot based on findings from biomechanics. The motor command of the robot is represented by time series data of muscle activation. The muscle activation patterns are determined from numerical calculation using a model of the musculoskeletal leg based on the measurement of muscle activity and kinetic data of the human movements. In the simulation results, the robot runs 8 steps with a speed of 3 m/s. We also demonstrate that the real bipedal robot is able to run for several steps.