Coordination of Mobile Agents for Simultaneous Coverage

摘要

Simultaneous environment coverage represents a challenging multi-agent application, in which mobile agents (drones) must cover surfaces by simultaneously capturing images from different viewpoints. It constitutes a complex optimization problem with potentially conflicting criteria, such as mission time and coverage quality, and requires dynamic coordination of agent tasks. In this paper, we introduce a decentralized coordination method, adaptive to a dynamic and a priori unknown 3D environment. Our approach selects the role an agent should take on and coordinates the assignment of agents to their computed viewpoints. Our main goal is to cover all detected objects in the environment at a certain quality as soon as possible. We evaluate the methods in AirSim in different setups and assess how the proposed methods respond to dynamic changes in the environment.