Sensorless Human-Robot Collaborative Assembly Considering Load and Friction Compensation

The collaboration of humans and robots can combine human flexibility with the efficiency of robots, so that robots can accomplish a variety of complex tasks, such as handling and assembly. However, most robots have poor collaboration capabilities and rely on expensive sensors to obtain the force interaction information of limited space range. To solve this problem, a human-robot collaborative assembly scheme without using additional force/torque sensors is proposed in this letter. First, a load identification and compensation method that does not require the load model is proposed to realize the decoupling of load force and human-robot interaction force. Next, to realize a compliant dragging effect of robots, an improved LuGre friction model is proposed based on the experimental discovery that the motion of the load has a non-negligible effect on the friction torque of robot joints. Finally, the kinesthetic teaching experiments of compliant assembly using a universal 6-DOF collaborative robot have been completed by designing a task-determined adaptive impedance control algorithm. The experimental results prove that the proposed scheme can effectively improve the safety, compliance, and flexibility of the loaded robots compared with the existing position-encoded robots, indicating its great potential for industry application.