Maxwell-Model-Based Compliance Control for Human–Robot Friendly Interaction

摘要

This article explores an alternative design philosophy of compliance control for safe and friendly human-robot interaction. Conventional approaches are based on the Voigt model with spring and damper connected in parallel. The theoretical and experimental results of disciplines other than robotics (e.g., mechanics of materials and elastoplastic mechanics) show that the Maxwell model has the merit of removing the elastic return force and reducing the effect of collisions with respect to the Voigt model. Therefore, the authors are motivated to exploit the advantages of the Maxwell model in reactions to human interactions and develop a series of Maxwell model-based compliance control methodologies, which only require kinematic control interfaces. First, a Maxwell model-based Cartesian admittance control scheme is presented to produce a novel plastic compliance of robot end effector. Moreover, we present a Maxwell model-based null space admittance control approach for manipulators with redundancy to achieve a novel plastic whole body compliance. We have evaluated the effectiveness and generalizability of the proposed methods by extensive comparative experiments on human-robot cooperative tasks in various sectors, including domestic, part assembly, and robotic puncture surgery. Our approaches can be viewed as a new and comfortable interface of physical human-robot interaction and collaboration.