Development of an Intelligent Controller for a Two-Axis Camera Vision Platform

Control systems designs based on single, classical algorithms can involve severe compromises among system response characteristics such as speed of response, relative stability and steady-state error. Additionally, they often are not very robust, and lack the capability to respond to changes in performance requirements. In practice, more desirable combinations of these characteristics can be obtained by using a mixture of control algorithms, based on both modern and classical theory, and carefully combining them so that the particular benefits of each are used to best advantage. This is one approach to intelligent control. This paper describes the ongoing development and testing of a two-level "intelligent" controller which applies this approach to elevation and azimuth position control of a camera pointing gimbal. The controller was designed using a detailed, non-linear computer simulation model and is implemented in a single Intel 80386/80387 based microcomputer. As predicted by the computer simulation model, and verified by the remarkably similar laboratory test data, the intelligent controller provides-when compared with a more classical controller earlier designed for the same hardware-a factor of 2 to 3 increase in bandwidth and a similar decrease in large-angle rise time, an order of magnitude decrease in overshoot for large-angle steps, and an order of magnitude decrease in steady-state error.