Toward a Heterogeneous Multi-robot Framework for Priority-Based Sanitization of Railway Stations.

We present a new framework for the prioritized multi-robot sanitization of railway stations based on Deep Reinforcement Learning. The proposed framework allows us to define teams of robots having different sanitizing strategies/capabilities, e.g., faster robots rapidly sanitizing small areas in cooperation with slower but long-range ones. Here, robot-specific policies are defined in order to accommodate the different capabilities of the single agents, while two global metrics are defined to assess the performance of the overall team. This capability of managing heterogeneous teams is an important requirement for the infrastructure manager Rete Ferroviaria Italiana S.p.A., which plans to verify to what extent different technologies or different strategies can be combined to reduce costs or increase cleaning efficiency. We tested our framework considering real data collected by the WiFi network of the main Italian railway station, Roma Termini, comparing its results with a similar Deep Reinforcement Learning system where homogeneous robots are employed.