A Comparative Experimental Study of Multi-Tasking Tracking and Interaction Control on a Torque-Controlled Humanoid Robot

Abstract

Multi-tasking control exploits kinematic redundancy of robots to attain several control objectives at the same time. To properly coordinate the subtasks according to their importance, they are usually stacked into a prioritized hierarchy. In this work, two passivity-based multi-tasking control strategies developed in our recent work that feature strict prioritization and mathematically proved stability properties, are experimentally compared with a state-of-the-art method using feedback linearization on a torque-controlled humanoid robot. The conducted experimental study aims at providing insights into the practical properties of the controllers in real-world scenarios whence the robot has to execute a mixture of trajectory tracking and physical interaction tasks.