Decentralized Coordination of Multi-Agent Systems Based on POMDPs and Consensus for Active Perception

This work presents the method based on the Partially Observable Markov Decision Processes (POMDP) and consensus protocol. The main idea is to share the belief and reach the consensus on the belief state in order to improve local decision making. To show that the belief update is important after reaching the observation, alongside the average consensus, we also examine novelty-biased consensus. The proposed method is applied on several benchmark problems and compared to an established method called Decentralized POMDP. Additionally, it is thoroughly examined in the simulation scenario. The results obtained in this work show that our method is efficient on the scenarios where agents explore the environment and it manages to execute mission in the scenarios where agents need to coordinate.