A Variable Admittance Control Strategy for Self-Collision Avoidance Based on Virtual constraints

Abstract

Collaborative robots are becoming assistants who support humans in completing various tasks in production and life. During human-robot collaborations, the robotic self-collision will cause damage to both operator and robot, and thus should be avoided. In this research, a control framework is proposed for solving the robotic self-collision problem based on virtual constraints and the adaptive admittance control. At first, an online algorithm is established to generate the virtual repulsive forces by considering the operator's intention and the distance between two robotic links. Then, the online algorithm is used to build a novel admittance controller for avoiding self-collision and achieving stability by using virtual forces and adjusting damping. Experiments are conducted on the AUBO i5F serial collaborative robot. Two other existing algorithms are employed to make comparison with the proposed one. It is revealed that the present control framework can not only effectively avoid the self-collision, but also have better comprehensive performance in enhancing the stability of the system and improving the comfortability of the collaboration.