Dynamic Perimeter Surveillance by Swarms of Small Robotic Vehicles

The goal of the proposed effort is the development and demonstration of an underwater plume monitoring tool. We propose to do this using a team of small underwater vehicles and an algorithm borrowed from the image processing literature. This system isexpected to provide vital support in three critical areas: drinking water supply surveillance, harmful algae blooms monitoring, and diver-based military missions.This effort will leverage an on-going university/industry collaboration between an image processing group (A. Bertozzi, UCLA) and an underwater vehicle research firm (Nekton Research, Durham, NC)). This collaboration has already produced workingmulti-vehicle algorithms for mine-countermeasures and for plume detection.The proposed system is based on a recent breakthrough in which an image processing technique called topological snakes was shown to be applicable to the multi-vehicle plume detection problem. The proposed effort will turn this breakthrough into a workingsystem. To this end, we will in Phase I provide a complete theoretical justification of the algorithm, including proofs of existence, uniqueness, and stability; examine critical issues related to its implementation on a live system.Since the algorithm is medium-independent, the findings of this effort are also applicable to teams of aerial or of terrestrial vehicles.In Phase II, we will perform a demonstration of the system using live vehicles. We will also design and test key upgrades, particularly: effect of number of vehicles; robustness under loss of robots and under limited communication bandwidth; stability improvements under low-bandwidth conditions; effect of plume dynamics; change in perimeter topology, i.e. fission and fusion; extension to three dimensional monitoring. This system is expected to provide vital support in three critical areas: drinking water supply surveillance, harmful algae blooms monitoring, and diver-based military missions.