Sensorless Collision Detection Method for Robots with Uncertain Dynamics Based on Fuzzy Logics

Abstract

Collison detection in human-robot collaboration plays an important role in avoiding the injuries to both humans and environment caused by robots. The traditional methods usually require additional sensors, which increase cost and the systematic complexity. And most sensorless methods assume that the robot dynamic model is accurate. The estimation of torque bounds based on fuzzy logics is proposed as an approach to detect collision in a sensorless robot with dynamic uncertainty. The dynamic model of the industrial robot is built and its parameters are identified firstly. Then the uncertainty of the dynamic model is analyzed and bounds of the actual torques are estimated online. The performance of the proposed collision detection method is evaluated using a 6-DOF industrial manipulator. The results show that collision can be reliably detected only with proprioceptive sensors.