Multiple autonomous vehicle mission planning and management

Abstract

This thesis investigates multiple autonomous vehicle mission planning and management. It begins by introducing the basic concepts and objectives of the multivehicle mission-planning problem. Then it formulates the problem mathematically and analyzes parameters in the objective function. The solution approach uses a hierarchical mission-planning scheme to take advantage of a scalable architecture. We develop a heuristic-based algorithm to solve the multiple-vehicle mission-planning problem. The algorithm has two phases: goal-point partitioning and routing. Goal-point partitioning uses a sweep procedure to group goal-points. Routing uses an implementation of simulated annealing combined with well-known TSP heuristics. Through the computational experiments conducted on both traveling salesman problem test cases, the TSPLIB library, and randomly generated test data, the routing algorithm performs quite well. It has been able to find TSP tours within one percent of optimality, and typically within one-half of one percent. The integration of the two-phase approach provides a solution to the multiple autonomous vehicle mission planning problem.