The development, analysis and evaluation of an optical tracker for tracking high-speed, manoeuvering targets

Abstract

This dissertation presents the design for an optical tracker which accurately quantifies the 1ine-of-sight between a target and the optical tracker. An optical tracker forms an essential element of defensive installations, yet publications on tracking schemes for optical trackers, as opposed to radar trackers, are hard to find. In this dissertation a comprehensive analysis of an optical tracker for high-speed, m a n o e u v e n n g targets has been d o n e . The optical tracker was to be mobile, and therefore subject to vehicle motion perturbations. A literature survey was m a d e , giving an overview of methoos available for tracking such targets and quantifying the resulting line-of-sight. The options considered included line-of-sight data obtained as outputs of Kalman or Linear Regression Filters of various orde r s . Tracking aids using linear filters, among o t h e r s , were also presented in the literature. The human operator was seen to be an integral part of the optical tracker. The literature survey pointed out some of the advantages inherent in including the human operator in series in the loop, but it was clear that difficulties would arise when trying to design an optimized controller around the human operator. The approach taken in this dissertation was to make a mathe= mat ical model of the system so that the controller could be analyzed, simulated and evaluated in the shortest possible time and at the lowest possible cost. This included human operator models a n d , in some cases, actual human operators. First the equations of motion for the gimballed platform containing the stabilized optical element were derived from first principles. Using the linearized version of these equations, classical controller deciqn methods (e.g. Root Locus techniques, Bode diagrams) were applied to determine the trans = fer function of the optimum stabilization control loop. The non-linear equations of motion were used to construct a simu= lator on which the controller could be tuned. The performance of the stabilization loops was found with respect to base motion stabilization as well as to input co m m a n d . Non-linearz