Human-Centred Design of an Automatic Train Regulation System

Modern urban light rail systems must deliver a frequent and regular train timetable in order to provide a good service to passengers. This paper describes how a human-centred design approach was taken to the provision of automated functionality that aims to augment human operators in one particular operational practice in the rail domain. Automatic Train Regulation (ATR) is a software tool that automatically modifies train departure times and coasting speed in order to optimise the service against timetable or headway. ATR achieves this by monitoring train movements and issuing control actions based on the detection of actual or predicted service disruptions. It was realised that the way that traffic controllers (signallers) and train drivers understand, interpret, and interact with the ATR system must be carefully designed from the outset. The challenge was to ensure that work practices informed the conceptual design of the automation, not just the visual appearance or interaction design. The paper describes how Human Factors methods were used to address these issues, and in particular how low fidelity prototyping and cognitive walkthroughs were used to explore the task-artefact cycle and raise user requirements during development of the system.