Train traffic control by re-planning in real-time

Today’s train traffic control systems are most often designed with a focus on controlling the technology rather than on to support the train traffic controller in achieving optimal traffic solutions. Train dispatchers often have to interact with many separate information systems in order to understand the traffic situation, to make decisions and to perform control tasks. An integrated and complete picture, showing the status of the entire traffic situation, is most often not at hand. We have found that train dispatchers create their own picture, in form of a dynamic mental model, based on long experience and available fragmented information. This is a very demanding cognitive process that forces the dispatcher to focus on understanding and gathering information about the situation instead on solving the problems. In the future these mental models will also be very difficult to maintain since new demands, more trains and higher speeds increase the complexity of the traffic process. Conflicts and disturbances also complicate the situation, forcing the train dispatcher to settle for a working rather than an optimal solution. However, by presenting all decision relevant information in a single graphical user interface and providing well designed tools for re-planning, we shift the focus from controlling the infrastructure to planning the traffic flow. By performing re-planning tasks in a computer based interactive time-distance graph the train dispatcher’s job will be to find a new optimal plan for the traffic. As long as this plan is free from conflicts, and the automatic train protection system is intact, it can be automatically executed by rather simple train route automates. This makes it possible for the train dispatcher to focus on the entire traffic situation and on solving conflicts as early as possible. To test our ideas we have developed an experimental work place, where prototypes of the new control system can be evaluated in a simulated traffic control environment.