Cooperative control and estimation for multiple agents with strongly coupled tasks

Cooperative control is still a fairly new field of study, where multiple autonomous agents work together to perform a task. We consider cooperative problems where the task causes strong coupling between the agents. In particular, we focus on an electronic attack problem. The goal is to use a team of unmanned air vehicles to deceive an enemy radar network by electronically generating a phantom radar track. To succeed, the vehicles must fly cooperatively and guide the phantom track so that each vehicle stays between one of the radars and the moving phantom. Significant challenges for this problem are the constraints on the vehicles and phantom, an information structure that provides flexibility for the vehicles, and accurate estimation of the radar positions.;We derive an analytic solution for locating a radar/transmitter based on time-differences of arrival. Using this solution and the Fisher Information Matrix, we show that there are simple rules for vehicle configurations that maximize estimation accuracy. We also use the analytic solution and Kalman Filtering to design estimation schemes, which are simulated. For online cooperative control of the phantom track and vehicles, we use optimal control theory to develop and apply practical methods for systems with state constraints. Finally, we use coordination structures to design a partially decentralized one-step algorithm for guidance, which is superior to existing ones as shown by simulations. A useful operational feature is that the algorithm allows dynamic replanning. A feasibility guarantee also predicts and corrects the one-step algorithm for constraint satisfaction.