Optimizing Multiagent Area Coverage Using Dynamic Global Potential Fields

摘要

Multiagent coverage algorithms have been used in the context of search and rescue operations to determine optimal search patterns for a team of robots. Many solutions to the problem of area coverage have been discussed in the literature. Our approach covers a physics-inspired technique for global control of agents in a search space. However, more importantly we adopt an evolutionary approach to evolve a policy for dynamically changing the global control rules driving a team of agents. We implement this algorithm using the Robot Operating System and the Gazebo simulation platform.