한국정밀공학회지 (Journal of the Korean Society for Precision Engineering)
- 제15권9호
- /
- Pages.93-100
- /
- 1998
- /
- 1225-9071(pISSN)
- /
- 2287-8769(eISSN)
비젼을 이용한 디버링 기술을 로봇에 전달
Transfer Deburring Skills to Robot Using Vision System
초록
This study presents the new method which can transfer the expert's skill to deburring robot through neural network. The expert's skill is expressed as association mapping between the characteristics of the burr and human expert's action. Under the fundamental idea that the state of the deburring process can be extracted via the visual sense of the human, we employ vision system for the perception and identification of the changing burr. From the demonstration of human experts, force data are measured. Finally the characteristics of the burr and corresponding force are associated by the neural network which is trained through many demonstrations. The proposed method is verified in the deburring process of welding burr.