Human-machine security collaboration based on virtual collision sensor

In this paper, a virtual collision sensor based on generalized momentum theorem is designed. The collision detection algorithm only needs to read the driving torque of the robot joint motor and the position information of the encoder to detect the collision that occurred during trajectory-following. Considering that the virtual collision sensor cannot effectively obtain the specific location information of the obstacles encountered by the robot arm, a collision response control strategy in trial progress inspired by the blind tactile detection environment is designed. The security mechanism of the robot is based on the master-slave task transformation algorithm. Finally, the virtual collision sensor and algorithm are simulated and verified by experiments in a 2D plane. The results show that the algorithm can achieve the goal of obstacle avoidance and trajectory tracking without the aid of external sensors and ensure the safety of man-machine cooperation. This algorithm is suitable for obstacle avoidance in an unknown multi-obstacle environment. It has the advantages of small computation and continuous change of avoidance speed.