Active vision tracking using an artificial intelligent control scheme

Abstract

Now most applications of vision guided robotic systems are in workspaces where the environments are strictly controlled. The vision system and the manipulator should be precisely calibrated before the system works. In this paper, a control structure combining an active vision system and a manipulator is proposed to simulate the mechanism of a man grasping a target. The binocular active eyes track the target and using inner sensors to simulate human's proprioceptor and give the information of the target pose relative to the "arm", namely, the manipulator. A kind of multistage fuzzy-neuro control strategy is adopted. The advantage of this method is that it can reduce the fuzzy rules number greatly. The simulations show that the system is stable and robust