Real time tracking in 3D space by robotic vision

A digital real time auto tracking robotic system is designed and analyzed. The tracking is based on passive detection using live TV imaging. The target coordinates are determined by an image processing technique and non-linear prediction algorithm. The control signals derived from the video signal set the azimuth ? and elevation f of the camera position to track the centroid of the target on the screen. The time development of the aim point in successive images represents the tracking path. Non-linear prediction determines the location of the target in the next TV frame by extrapolation. For distant objects a range finder can be mounted with the camera to determine the coordinate, r. Hence, a set of spherical coordinates (r, ?, f) for the target are determined