A Frequency Domain Approach for Contact Type Distinction in Human–Robot Collaboration

This letter addresses the problem of distinguishing unexpected collisions from voluntary contacts during human-robot collaboration. The objective of this letter is to improve the safety of the operator during the interaction with the robot. Based on the observation that cooperation forces present different frequency characteristics from collisions, a novel model-free method is developed to distinguish contacts using the spectral derivative of the external forces measured by the robot's proprioceptive sensors. In an experimental comparison with other approaches appeared in the literature, the proposed method presents faster distinction time, in various collaborative or involuntary contacts with multiple subjects.