Analyze assembly skills using a motion simulator

Abstract

Many researchers have focused on transferring human skills to robots by extracting those skills into the machine- understandable models. It is an important step towards creating an intelligent robot in a cooperative environment. These models not only can be employed to transfer human control strategy to robots, but also can be used to improve the new workers' performance. In this paper, we propose a method that is to extract the advantages of several operators' skills and synthesize new skills to train new workers. To realize the skill extract, synthesis and transfer system, it is absolutely necessary to reproduce and synthesize the experts' motions. In the latter case, a kinematic simulator of human body is employed to realize the synthesis of the operators' motions based on the detected motion data. To verify the proposed method, we employed it to execute a peg-in-hole assembly task. The results show that after training by the synthesized skill model, the new operator's performance is improved significantly.