(U) Dynamic Distributed Cooperative Control of Multiple Heterogeneous Resources

This research is concerned with dynamically determining appropriate flight patterns for a set of autonomous UAVs in an urban environment, with multiple mission goals. The UAVs are tasked with searching the urban region for targets of interest, and tracking those targets that have been detected. We assume that there are limited communication capabilities between the UAVs, and that there exist possible line of sight constraints between the UAVs and the targets. Each UAV (i) operates its own dynamic feedback loop, in a receding horizon framework, incorporating local information (from the perspective of UAV i) as well as remote information (from the perspective of the ‘neighbor’ UAVs) to determine the tasks to perform and the optimal trajectory of UAV i (and neighbor UAVs) over the planning horizon. This results in a decentralized and more realistic model of the real-world situation. As the coupled task assignment and flight route optimization formulation is NP-hard, a hybrid heuristic for continuous global optimization is developed to solve for the flight plan and tasking over the planning horizon. Metrics capturing the price of anarchy and price of decentralization are developed, and experimental results are discussed.