Distributed Target Detection and Coverage for Holonomic UAVs

Abstract

This thesis implements a novel distributed, deterministic algorithm for as few unmanned agents as possible to detect and cover as many static targets of unknown location as possible. This algorithm, Pruning-Perception-Decision (PPD), strikes the balance of exploration versus exploitation by maximizing the number of targets covered by each agent. Agents can cover only one grouping of targets at a time and continue exploring until they find an uncovered target. In doing so, agents' search area is discretized into a grid, where the average percent coverage of each tile is monitored with respect to each agent's field-of-view. Once all agents are covering targets and the average area-coverage value stabilizes, PPD terminates. Alternately, if all targets are found and there exist additional explorer agents, PPD terminates when a time threshold is reached. Simulations show that implementing PPD results in faster convergence than the state-of-the-art by nearly an order of magnitude as well as improved target coverage. Additionally, results of a second demonstration suggest that PPD could be applied to targets appearing and disappearing.