An Adaptive Approach to Vision-based Formation Control

Abstract

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. However, the lack of information regarding the state of motion of neighboring vehicles can lead to degraded performance and even instability. This paper presents an adaptive output feedback approach for addressing this problem. We design adaptive formation controllers that allow each vehicle in formation to maintain separation and relative orientation with respect to neighboring vehicles, while avoiding obstacles. The method works by enabling each vehicle in the formation to adaptively correct for the effect that the motions of neighboring vehicles have when regulating relative variables like range and line of sight. It is assumed that estimates of these variables can be derived using passive, vision-based sensors. The need for explicit communication to maintain formation is minimized and the resulting controller solution is decentralized. We implement a reactive obstacle avoidance controller to navigate in an environment with obstacles. The formation controller and obstacle avoidance controller are outer-loop controllers whose outputs are speed and heading commands. These commands are blended together to generate composite speed and heading commands that are inputs to the inner-loop controller. The weights used for blending the commands depend upon the priority of the task at hand. We illustrate the method with an example involving a team of three aircraft keeping formation in the presence of obstacles.