Robust Visual Tracking Control of Pan-tilt -Stereo Camera System

Abstract

In this paper, a new virtual 3D stereo camera model is presented to get a full rank of the image Jacobian matrix (Jimg) for an eye-in-hand stereo camera system. Some problems of the classical image Jacobian are local minima and singularities in image space solving by Jimg. Furthermore, a robust visual control method is also proposed for a pan-tilt platform with stereo camera-in-hand using a neural network to compensate the effect of uncertainties in the dynamics of the overall system. An online learning neural network is used in the closed loop control system to compensate the uncertainties of both the Jacobian matrix and the dynamics of pan-tilt robot. The asymptotic stability of the overall visual control system is proved by Lyapunov stability method. The performance of the proposed control method is demonstrated by simulations in Matlab.