A neural visual servoing in uncalibrated environments for robotic manipulators

In this paper we describe an image based approach for the visual control of robotic manipulators, which uses neural networks to cope with calibration inaccuracies and relevant changes in the geometry of the system. A fast sliding-mode based algorithm has been employed for the on-line training of three neural networks approximating the relationship between camera coordinates and world coordinates. The proposed approach is tested on the simulations on a 5-dof robotic manipulator that must track a moving object using a stand-alone stereoscopic vision system.