Using Control And Vision For Space Applications

Abstract

In this paper, we present algorithms that address the real-time robotic visual tracking (eye-in-hand configuration) of satellites that move in 2-D space. To achieve our objective, we combine computer vision techniques, for detection of motion, with simple control strategies. The problem has been formulated from the systems theory point of view. This facilitates the extension of the algorithm to 3-D space. A cross-correlation technique (SSD optical flow) is used for computing the vector of discrete displacements and is combined with an appropriate control scheme to calculate the required motion of the space robotic system. Shared and traded control modes enable the integration of the human operator with the autonomous tracling modules. In this way, the human operator can intervene and correct the tracking motion through a joystick. The performance of the proposed algorithms has been tested on a real system, the CMU DD Arm 11, and the results are presented in this paper.