Active vision-based control schemes for autonomous navigation tasks

摘要: Abstract This paper deals with active-vision-based practical control schemes for collision avoidance as well maintenance of clearance in a-priori unknown textured environments. These employ a visual motion cue, called the threat cue (VTC) sensory feedback signal to accomplish desired tasks. The VTC provides some measure relative change range between 3D surface and moving observer. It is collective obtained directly from raw data gray level images, independent type texture. measured [time−1] units needs no reconstruction. are based on set If–Then fuzzy rules almost knowledge about vehicle dynamics, speed, heading direction, etc. They were implemented real-time using 486-based Personal Computer camera capable undergoing 6-DOF motion.