Recognition and Relocation of 3-Dimensional Objects Using Stereo Robot Vision System

An important task in sterco robot vision is to determine the three-dimcnsional locations of objects. The real location of an object is determined from the disparity of matched pints in sterco image pairs. This papcr dcpicts a new sterw correspondence techniquc to match planes extracted from stcreo image pairs. The stereo image pairs are taken from the lateral stereo camera model of eye-on-hand configuration with a camera on the end-cffcctor of the robot. Stcreo Robot Vision Systcm (SRVS) which finds out thc three-dimensional locations of objects is described. Its software architecture consists of five major steps: stereo camcra modclling, feature extraction, sterco image matching, depth dctcrmination, and interpolation. In ordcr to realize a knowledge-bascd robot which has the capabilities of cognition and problem solving, we make the robot recognize objects and rclocate them in the spccificd order using the three-dimensional locations obtaincd from the SRVS. Wc show results of the SR...

• Type: Conference
• Year: 1990
• Where: MVA
• Authors: Kyoung Mihn Do, Chee-Woo Kang, Kwae-Hi Lee
• Keywords: Image Pairs, MVA 1990, MVA 2007, Robot Vision, Sterco Robot Vision