ADAPTIVE ROBOTIC VISUAL TRACKING - THEORY AND EXPERIMENTS

This paper addresses the use of a vision sensor in the feedback loop w ithin the Controlled Active Vision framework. Using this framework, al gorithms are proposed for the solution of the robotic (eye-in-hand con figuration) visual tracking and servoing problem. We state the problem of visual tracking as a problem of combining control with computer vi sion. We use the sum-of-squared differences (SSD) optical flow for the computation of the vector of discrete displacements. The measurements can be derived either from a single big window or from multiple small windows. These displacements are fed to an adaptive controller (self- tuning regulator) that creates commands for a robot control system. Th e whole algorithm is based on the on-line estimation of the relative d istance of the target with respect to the camera. An important contrib ution of this work is that it requires only partial knowledge of the r elative distance of the target with respect to the camera. This fact o bviates the need for off-line calibration of the eye-in-hand robotic s ystem. We have implemented, both in simulation and in experiments, thr ee different adaptive control schemes, and the results are presented i n this paper. The computational complexity and the experimental result s demonstrate that the proposed algorithms can be implemented in real time.