A Heuristic Method for Task Selection in Persistent ISR Missions Using Autonomous Unmanned Aerial Vehicles
Date of Award
Doctor of Philosophy (PhD)
Department of Aeronautics and Astronautics
Donald L. Kunz, PhD
The Persistent Intelligence, Surveillance, and Reconnaissance (PISR) problem seeks to provide timely collection and delivery of data from prioritized ISR tasks using an autonomous Unmanned Aerial Vehicle (UAV). In the literature, PISR is classified as a type of Vehicle Routing Problem (VRP), often called by other names such as persistent monitoring, persistent surveillance, and patrolling. The objective of PISR is to minimize the weighted revisit time to each task (called weighted latency) using an optimal task selection algorithm. In this research, we utilize the average weighted latency as our performance metric and investigate a method for task selection called the Maximal Distance Discounted and Weighted Revisit Period (MD2WRP) utility function. The MD2WRP function is a heuristic method of task selection that uses n+1 parameters, where n is the number of PISR tasks. We develop a two-step optimization method for the MD2WRP parameters to deliver optimal latency performance for any given task configuration, which accommodates both single and multi-vehicle scenarios. To validate our optimization method, we compare the performance of MD2WRP to common Traveling Salesman Problem (TSP) methods for PISR using different task configurations. We find that the optimized MD2WRP function is competitive with the TSP methods, and that MD2WRP often results in steady-state task visit sequences that are equivalent to the TSP solution for a single vehicle. We also compare MD2WRP to other utility methods from the literature, finding thatMD2WRP performs on par with or better than these other methods even when optimizing only one of its n + 1 parameters. To address real-world operational factors, we test MD2WRP with Dubins constraints, no-y zones in the operational area, return-to-base requirements, and the addition and removal of vehicles and tasks mid-mission. For each operational factor, we demonstrate its effect on PISR task selections using MD2WRP and how MD2WRP needs to be modified, if at all, to compensate. Finally, we make practical suggestions about implementing MD2WRP for flight testing, outline potential areas for future study, and offer recommendations about the conduct of PISR missions in general.
DTIC Accession Number
Olsen, Christopher C., "A Heuristic Method for Task Selection in Persistent ISR Missions Using Autonomous Unmanned Aerial Vehicles" (2018). Theses and Dissertations. 4421.