Subtask-learning Based for Robot Self-Assembly in Flexible Collaborative Assembly in Manufacturing
Zhang Songsong,Huang Haisong,Huang Dong,Yao Liguo,Wei Jianan,Fan Qingsong
DOI: https://doi.org/10.1007/s00170-022-09177-1
IF: 3.563
2022-01-01
The International Journal of Advanced Manufacturing Technology
Abstract:In recent years, human-robot collaboration has been introduced into production systems as a new generation of production modes, and robot learning technology based on human demonstrations has become increasingly suitable for industrial manufacturing scenarios. This paper proposes a robot learning framework based on human demonstration. Through human demonstration, the robot learns the part grabbing, assembly sequence, assembly path, and assembly actions in assembly tasks. In this framework, for actual production scenarios in industrial applications, a robot learning grasping method based on an improved Gaussian model is adopted, and humans can complete high-precision grasping tasks only through teaching. In addition, in the human-robot collaborative assembly process, the Gaussian mixture model and improved RPD algorithm are used to perfect the robot task learning, so that the robot can successfully complete the assembly task.