Cooperative multi-vehicle search and coverage problem in uncertain environments

A decentralized approach is proposed to solve a cooperative multi-vehicle search and coverage problem in uncertain environments. Two different types of vehicle are used for search and coverage tasks. The search vehicles have a priori probability maps of targets in the environment and they update these maps based on the measurement of their sensors during the search mission. They use a limited look-ahead dynamic programming algorithm to find their own path individually while their objective is to maximize the amount of information gathered by the whole team. The task of service vehicles is to spread out over the environment to optimally cover the terrain. A locational optimization technique is used to assign Voronoi regions to vehicles and the stability of coverage system is guaranteed using LaSalle´s invariance principle. The service vehicles modify their configuration using the updated probability maps which are provided by the search vehicles. Simulations show that the proposed approach offers improved performance compared to conventional coverage methods.