Trajectory Imitation With Visual Guidance and Compliance Control for Robotic Bolt Grasping and Assembly

Abstract

Robotic bolt grasping and assembly holds significant practical value in automated manufacturing settings. In this article, a three-stage framework for bolt grasping and assembly is proposed, which includes a trajectory learning stage, a bolt grasping and arranging stage, and a bolt assembly stage. In the trajectory learning stage, the demonstration trajectories for bolt grasping and bolt arranging are encoded using dynamical movement primitives (DMPs). In the bolt grasping and arranging stage, DMPs with visual guidance are proposed to grasp a horizontally positioned bolt and vertically place it on an assembly board. In the bolt assembly stage, a linear and sinusoidal motion planning approach with force control along the determined rough direction is presented for hole search. Then screw motion with compliance control is employed for the bolt installation. Finally, robotic grasping and assembly experiments are conducted for different bolt sizes and initial poses. A series of experimental results verify the effectiveness of the proposed robotic bolt grasping and arranging method, as well as the linear and sinusoidal hole search approach. The proposed framework can significantly improve the accuracy and efficiency of robotic bolt grasping and assembly.