UNCALIBRATED 3D ROBOTIC VISUAL TRACKING BASED ON STEREO VISION

This paper proposes a new nonlinear visual mapping model for the 3D ro botic visual tracking problem and a novel visual tracking controller is designed based on artificial neural network without explicit external and internal calib ration. Simulation results show that this method can drive the static tracking e rror to zero quickly and keep good robustness and adaptability at the same time. Additionally, the algorithm is very easy to be implemented with low computation al complexity.