LSTM Path-Maker: a new LSTM-based strategy for the multi-agent patrolling

AbstractFor over a decade, the multi-agent patrol task has received a growing attention from the multi-agent community due to its wide range of potential applications. However, the existing patrolling-specific algorithms based on deep learning algorithms are still in preliminary stages. In this paper, we propose to integrate a recurrent neural network as part of a multi-agent patrolling strategy. Hence we proposed a formal model of an LSTM-based agent strategy named LSTM Path Maker. The LSTM network is trained over simulation traces of a coordinated strategy, then embedded on each agent of the new strategy to patrol efficiently without communicating. Finally this new LSTM-based strategy is evaluated in simulation and compared with two representative strategies: a coordinated one and a reactive one. Preliminary results indicate that the proposed strategy is better than the reactive.

Return to previous page