A robotic peg-in-hole assembly method based on demonstration learning and adaptive impedance control

Purpose This study aims to address the issues of complex modeling and weak adaptability to environmental changes in traditional robotic peg-in-hole assembly methods, a new peg-in-hole assembly approach based on demonstration learning and adaptive impedance control is proposed. Design/methodology/approach First, this paper developed an overall assembly strategy by performing force and geometric analyses of the peg during the assembly process. Then, using demonstration learning, this paper enabled the robot to learn force information specific to the insertion process. Finally, this paper proposed an adaptive impedance controller to track the desired force in unknown environments, ensuring the stability of the robot’s assembly operations. Findings Experimental results demonstrate that the proposed method exhibits strong robustness to both peg-hole clearance and hole positioning errors. Ten repeated experiments were conducted for each of the three different clearance sizes, all successfully completing the assembly. In addition, the average assembly time was under 20 s, highlighting the efficiency and reliability of the method. Originality/value This provides a novel approach for robotic peg-in-hole assembly tasks. This method eliminates the need for complex physical modeling while offering high robustness to positioning errors and variations in peg-hole clearance.