(Talk Cancelled)

Virtual coupling is the closer running of trains where they are only separated by their relative braking distance plus a factor of safety, so that the following train shall not collide with the leading train if the emergency brakes are applied. This project aimed to investigate the potential capacity increase a virtual coupling approach offers the railway in Britain and determine its technical feasibility. This presentation is based off the presenter’s undergraduate research project and will focus on:

• The need for increased capacity on the railway,

• The components of a virtual coupling system,

• Computer modelling of a virtually coupled system and its applications,

• Simulation results and potential capacity increases,

• Project limitations, assumptions, and achievements.

The Society's Annual General Meeting will be followed by a talk on 30 years of line signalling upgrades

The talk will reflect on on the 30 years of whole line signalling upgrades on TFL lines. Starting with the Central line in 1980’s the presenter will talk us through the salient events that beset the DLR, Jubilee line extension, Jubilee, Northern, Victoria and more recently sub surface railways line upgrades. Building on commentary senior managers in TfL and independent reviews done on the performance of the upgrades reflections will be offered on what lessons we can take out of this for the future.

Rapidly evolving artificial intelligence technology from the autonomous car industry offers the opportunity for the rail industry to improve the safety, efficiency and cost-effectiveness of on-board driver advisory systems, next-generation train control systems, and rail systems. The presentation will describe our progress in applying artificial intelligence and machine learning technologies to rail operations in Australia and overseas. The specific topics of hazard and infrastructure detection, and authority limits monitoring and braking, will be discussed. I will summarise with some observations on where AI is compelling today.

For safe and efficient operations, a rail vehicle needs to know exactly where it is, recognise the hazards around it, and continuously monitor its authorised route for both normal and abnormal operations. AI can undertake this role by integrating multiple 24/7 sensor data in real-time to allow a comparison with the previous record of any given track section. It does this by using neural network processing in an on-board processor to interpret real-time imagery and other sensor data. Software processes undertake the computationally intensive statistical algorithms for localisation, object detection, awareness, dynamics and route monitoring, and ultimately alarming.