High Performance Torque Control of Drive Systems for Legged Robots

Abstract

This paper focuses on the design of high performance torque control drive system for legged robots. Achieving robust torque response performance is essential for the motors employed in these robots. Considering different transmission configurations, we introduce two distinct inertia models: one-mass system and two-mass system. For the case of a one-mass system, we design a current loop control strategy based on the Dead-Beat Control (DBC), which achieves faster torque response. In the context of the two-mass system, we propose a Model Predictive Control (MPC) method to effectively enforce constraints on both electromagnetic torque and shaft torque. Subsequently, we conduct separate experimental validations for these two proposed models, leading to superior performance as compared to traditional methods.