An adaptive vision-based approach to decentralized formation control.pdfVIP

JOURNAL OF AEROSPACE COMPUTING, INFORMATION, AND COMMUNICATION Vol. 1, December 2004 An Adaptive Vision-Based Approach to Decentralized Formation Control Ramachandra Sattigeri * and Anthony J. Calise? Georgia Institute of Technology , A tlanta, Georg ia 30332-0150 and Johnny H. Evers ? A ir Force Research Laboratory, Eglin A ir Force Base, F lorida 32542-6810 In considering the problem of formation control in the deployment of intelligent munitions, it would be highly desirable, both from a mission and a cost perspective, to limit the information that is transmitted between vehicles in formation. We have proposed an adaptive output feedback approach to address this problem. Adaptive formation controllers are designed that allow each vehicle in formation to maintain separation and relative orientation with respect to neighboring vehicles, while avoiding obstacles. We have implemented two approaches for formation control, namely, leader-follower formations and leaderless formations. In leader-follower formations, there is a unique leader and all the other vehicles are followers. In leaderless formations, there is no unique leader. Each vehicle tracks line-of-sight range to up to two nearest vehicles while simultaneously navigating towards a common set of waypoints. As our results show, such leaderless formations can perform maneuvers like splitting to go around obstacles, rejoining after negotiating the obstacles, and changing into line-shaped formation in order to move through narrow corridors. Nomenclature UAV unmanned aerial vehicle MAV micro autonomous vehicle LOS line-of-sight NN neural network MIMO multi-input multi-output DOF degree of freedom PCH Pseudo-Control Hedging SHL Single Hidden Layer IMU Inertial Measuring Unit LDC Linear Dynamic Compensator WP Waypoint N V Number of Vehicles Tracked cmd command com c