Deep and Decentralized Multiagent Coverage of a Target With Unknown Distribution

This article proposes a new architecture for multiagent systems to cover an unknown distributed target quickly and safely and in a decentralized manner. The interagent communication is organized by a directed graph with a fixed topology. The author models agent coordination as a decentralized leader–follower problem with time-varying communication weights. Given this problem setting, the author first presents a method for converting the communication graph into a neural network, where an agent can be represented by a unique node of the communication graph but multiple neurons of the corresponding neural network. The author then applies a mass-centric strategy to train time-varying communication weights of the neural network in a decentralized fashion. This implies that the observation zone of every follower agent is independently assigned by the follower based on positions of its in-neighbors. By training the neural network, the author can ensure safe and decentralized multiagent coverage control. Despite the target is unknown to the agent team, the author provides a proof for convergence of the proposed multiagent coverage method. The functionality of the proposed method is validated by a large-scale multicopter team covering distributed targets on the ground.