Robust Vision-Based Target Tracking Control System for an Unmanned Helicopter Using Feature Fusion

We present in this paper a robust vision-based tracking control system for an unmanned helicopter to track a moving ground target. It integrates a real-time vision-based target detection algorithm with a tracking control law in a closed loop. First, the proposed target detection algorithm extracts geometry, color and motion features from captured images. Based on these features, a finite-state machine is then introduced to dynamically coordinates the work of decision making under the Bayes framework, and a tracking control law is designed to minimize a certain tracking error function. Experimental results obtained from actual flight tests are also presented and demonstrate the effectiveness and robustness of our vision-based tracking control system in real scenes.