Design of a stereo robot head: hardware and software issues

Current research in machine vision encompasses a wide variety of systems developed with a broad spectrum of characteristics. Many of these existing systems are prohibitively expensive or overly complicated for implementation in an undergraduate research program. This paper presents a stereo vision system designed to be low-cost, light-weight, and compact. The system is intended for use in extracting 3-D surface information from a scene and providing this information to a recognition system.