Determining Grasp Points using Photometric Stereo and the FKISh 4 Bi ~ iocular Stereo System

Katsushi Ikeuchi,H. Keith Nishihara, P. UertholtiK., I-Jorn,Patrick Sobalvarro, Shigerni Nagata

msra(2004)

引用 22|浏览37
暂无评分
摘要
This paper describes a system which locates and gra:zps doughnut shaped parts frcrn n pilc. The system uses photometric stereo and binocular slcreo as vision input to&. Phot,onietric stereo is used to make surface orientation measur~ments . With chis ini'ormstion t h e camers Geld is segmented into isolated regions of' continuous sn100th surface. One of these regions is then sclcctcci as the target rcgion. The a!,tii,lide of the physical object associated with thc target rcijori is dcterrnined by histograming surface orientations over Lhat regioll nnd comparing with stored histograms obtained From proto+,ypical objects, Range information, riot available from photometric stereo is 0bt~1ined by the PRISM bir:ocul:ir stereo system. A collision-free grasp configuration and approach trajectory is con~putec! 2nd executed using the att i tude, and range data.
更多
查看译文
关键词
photometric stereo,lcgal grasp congguration,robot vision.,din-picking,binocular stereo,extended gaussian inlase
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要