Establishing Trajectories for Multi-Vehicle Reconnaissance

Abstract

A methodology utilizing multiple, cooperating vehicles for reconnaissance is developed. Teams of vehicles can drastically improve performance (time, information/uncertainty) by sharing information among vehicles and performing cooperative estimation. Given a set of targets and vehicles, there is a need to plan both the trajectories and task assignments together such that vehicles gather as much information on the targets as possible. The paper presents a strategy for developing task assignment and rough reconnaissance trajectories for two or more vehicles. The work builds on optimal planning results for simpler problems, using them as a foundation for heuristics. Additional work on clustering and cost bounding are used to alleviate the computational load, while maintaining performance. The methodology is shown to recover optimal results for over 90% for the cases up to five targets. Research using cooperating multiple vehicles has been widely explored in recent years. By using all resources through cooperation among vehicles in a team, the overall performance can be improved. Applications that take advantage of this performance increase using multi-vehicles include survey/mapping in an unknown environment, battlefield attack, and defense. Designing cooperative control methodologies of multiple vehicle systems for these tasks can be challenging because the size and numerical complexity of the problem become significantly larger as the number of vehicles grows. It is desirable to maintain performance without undue computational burden as the problem size increases. In Ref. 1, cooperative control of multiple UAV’s performing series of tasks on targets is investigated. A network flow and auction algorithm are used in the multiple assignment problem, which the environment assumes no obstacles. The work here focuses more specially on cooperative reconnaissance, with a more thorough study as to why the vehicles cooperate and how performance increases. Coordinated control of multiple robot is addressed in Refs. 2, 3 using an information based approach. This work nicely makes use of information theory to guide the vehicles over paths using a numerical solution to the optimal control problem, and then decentralized data fusion algorithm for estimation. Subsequent work on the scalability of the proposed approach to larger numbers of targets and vehicles, obstacles and moving targets are still open questions. In Ref. 4, task assignment for multiple vehicles in the presence of polygon obstacles is explored using Mixed Integer Linear Programming. Optimal solution for weapon target task assignment is shown in Ref. 5, and approximate nearly optimal, less computationally intensive solution has also been found in Ref. 6. The problem involving multiple observers and targets is solved in real time using a heuristic approach in Ref. 7. The objective of this work is to develop a methodology for cooperative reconnaissance using a hierarchical control architecture (i.e., team path planning/task assignment) on a cooperative multiple vehicle system. Specially, given a set of targets with varying levels of priorities and uncertain locations (a prior), the methodology plans task assignments and paths for the multiple cooperating vehicles that maximize performance (information/uncertainty as a function of time). Given a set of targets with varying levels of priorities and uncertain locations (a prior), the methodology plans task assignments and paths for the multiple cooperating vehicles that maximize performance (time, ¤ Research Assistant, student member AIAA †Associate Professor, senior member AIAA