摘要: Many navigation tasks, such as the use of unmanned vehicles for planetary exploration or defense reconnaissance, require onboard range sensors to automatically detect obstacles in path a vehicle. Passive ranging via stereo triangulation, vision, is very attractive approach obstacle detection because it nonemissive, nonmechanical, nonscanning, and compatible with stereographic viewing by human operators. However, several problems have restricted practicality past, including limitations on speed, reliability, generality existing matching algorithms. This situated has changed, Jet Propulsion Laboratory (JPL) recently demonstrated first semi-autonomous, robotic traverses natural terrain vision detection, all computing article reviews main algorithmic paradigms describes near realtime system developed at JPL, presents experimental results that demonstrate emerging practically semi-autonomous land navigation. © 1992 John Wiley & Sons, Inc.