Connected or Autonomous Trains?

In a parallel with the trends in the automotive domain, we discuss the future challenges of automation of train control, where train to infrastructure and train to train communication will support distributed control algorithms, while on board artificial intelligence will provide autonomous control decisions. Already installed systems, like ERTMS-ETCS, are actually distributed systems that span over geographical areas and are able to safely control large physical systems. But still, crucial decisions are taken at centralized places, that concentrate communications with mobile objects. Several prospected advances, aimed at increasing capacity and automation of rail transport, go in the direction of a more dynamic network connection among mobile components, in which decisions are actually taken in a distributed way. A concept of dynamic safety envelope within which a train can safely move then emerges, built by a fusion of reliable information coming from the infrastructure and from other trains, as well as autonomously harvested by on-board “intelligent” sensors. This paper discusses some plausible scenarios in this respect and presents the basic concepts behind them.