Research on Collision Detection of Collaborative Robot using improved Momentum-based Observer

Abstract

With the increasing application of collaborative robots, it is essential to ensure staff safety during human-machine collaboration, so the robot collision detection problem must be addressed. Traditional collision detection requires external sensors, which increases the cost and the structure's complexity. This paper constructed a generalized momentum-based observer to detect external collisions and introduced a compensation link and variable damping design to improve collision detection speed and accuracy. At the same time, after witnessing an external collision, a collision response strategy based on time-scaling was adopted to protect staff safety. Through algorithm simulation, it was verified that the improved momentum-based observer had significantly improved the real-time performance and detection accuracy. At last, the proposed collision detection algorithm was carried out on a collaborative robot with 6 degrees of freedom. The results show that the proposed collision detection algorithm can accurately detect external collisions without external sensors and effectively protect staff safety.