Robots on rails
Autonomous trains are coming. We explore how our customers could benefit from this revolutionary technology – and where it might ultimately take us.
Why do we need autonomous trains?
With the world’s population rapidly growing - and more and more people living in cities, mobility has become a one of the toughest challenges, in terms of quality, safety and sustainability.
Railway operators believe that automation is the future of the train. Autonomous train can bring much more in solving mobility problems, by increasing the network capacity and therefore moving more people and goods, more regularly, and also reducing energy consumption by optimizing driving. Autonomous trains are going to be smart, connected trains that meet both cost and quality of service issues.
Railways have long experience with automation: driverless metros have been a reality for more than 30 years. They also have fewer moving vehicles and there is less infrastructure to monitor.
Autonomous operation is a golden opportunity for rail. We believe that autonomous trains – trains that can “see” and “think” – will provide operators with a decisive competitive advantage, not only by reducing costs, but also by transforming the capacity, flexibility and attractiveness of their networks.
Will robots change the way we operate railways?
We believe that autonomous trains will have a tremendous impact on the way railways are run.
The business case for autonomy rests primarily on factors such as extra capacity, improved flexibility (including rapid recovery from disruption) and reduced operating costs. Autonomous technology also has the potential to improve safety by detecting obstacles that would not normally be visible to a human driver.
The value of these capabilities is enormous. But autonomy goes beyond simply improving the status quo. We believe that autonomous technology will allow our customers to envisage entirely new ways of using their infrastructure.
One early-stage example is depot automation. Autonomous trains can use their sensors to support “close-up” and couple operations without the need for central control and co-ordination. This not only allows full depot automation, but also has the potential to reduce land take for depots because trains make the best use of available space.
The fact that trains can “see” makes them a potential source of survey data. This opens the door to a far deeper understanding of the condition of the infrastructure than is currently possible with periodic surveys and track inspections.
The ability to stop trains accurately at stations – another characteristic of autonomy – also has interesting potential. In combination with passenger communications and selective door control, for example, it would be possible for operators to provide much shorter platforms at lightly-used stations. This would not only reduce maintenance costs, but also make it cheaper to build new stations. It would even allow operators to experiment with “pop-up” stations to cater for specific events.
Autonomy is also an enabler for “virtual coupling” – the ability for individual trains to travel safely in convoy. This has benefits both in terms of capacity and energy savings. Automotive manufacturers have been working on a similar technology for lorries for a number of years.
What impact will autonomy have at the trackside?
Robots are capable of “seeing” their environment and they will be capable of reading and responding to lineside signals. This means that our customers will be able to benefit from autonomy without changing their signalling systems. This applies to both metro and main line operations. Mixed-mode operation will also be feasible.
In the longer term, as intelligence migrates from track to train, the need for trackside infrastructure will be progressively reduced. Lineside signals, signs and warning boards will no longer be needed. The biggest change, however, will be in the shift to train detection based on self-localisation: on main lines, for example, trains will use satellite positioning and inertial measurements to determine their position, instead of track circuits or axle counters.
There are clearly savings to be realised from the reduction in trackside systems. However, there is a judgement to be made about the extent to which the track should be allowed to become “dumb”. There is a case for maintaining a dialogue between track, train and control centre. Fibre optic sensors, for example, are already coming into use for train detection. The secondary functions offered by this technology – measurements of speed, weight and wheel condition – have enormous potential value for operators.
What about the impact of robots in the wider world?
This is an important consideration. As a rising tide carries all ships, so the technology that makes robotic trains possible will also cause revolutionary changes across the transport sector and beyond.
On one hand, these changes could be an existential threat to railways. What if self-driving cars became so cheap that people stopped using trains? Both ride sharing and car autonomy are closely linked – ride-hailing fleets will provide the first major opportunity for full autonomy. Consulting firm BCG estimates that nearly a quarter of all car passenger miles travelled in the United States will be in shared autonomous electric vehicles by 2030.
Yet there are two parallel developments that suggest the future relationship between roads and railways will be collaborative rather than destructive. The first is that our relationship with cars is changing – we’re increasingly likely to lease or share vehicles, rather than owning them outright. And young people are driving less. In short, our attachment to cars is weakening.
The second factor is the rise of Mobility-as-a-Service (MaaS) and journey planning apps. These shift the focus from the mode to the journey. End-to-end journeys will be assembled using rational combinations of rail and road components based on consumer-defined parameters, such as speed or cost. If rail is able to compete on these fronts, then the future is one of greater opportunities as part of a genuinely multimodal transport ecosystem.
The rise of autonomous delivery networks could also favour railways. Indeed, the deeper integration of railways into the wider multimodal ecosystem – a shift which is implicit in both the rise of autonomy and the huge potential for digitally-arbitraged goods transportation – could herald the revival of railway participation in the door-to-door freight market.
Finally, what new challenges are envisaged?
As railways run closer to their theoretical maximum capacity, new challenges could emerge. Rising traffic levels will amplify reactionary delays, so greater flexibility will be required. Wear and tear on the track will increase, so new approaches to maintenance will be needed. Meanwhile, greater passenger flows could create capacity bottlenecks on platforms and stations.
Some of the solutions to these problems are inherent in autonomous technology itself. For example, trains themselves will be capable of conflict resolution and able to operate bi-directionally. It is also possible that parallel developments in MaaS will lead to higher levels of synchronisation between different modes of transport. This might, for example, lead to lower levels of congestion on stations and platforms by reducing waiting times.
However, the ability to orchestrate operations will be critical. This requires a full architecture that takes account of both the autonomous train fleet and the network over which it operates.
Traffic Management Systems will play a pivotal role here and predictive capabilities will be of fundamental importance. Traffic management will not only need to take care of the network itself, but will also need to take into account all the external factors that influence it – everything from weather and large public events, to incoming data from MaaS apps about individual travel intentions. To get the full benefits of autonomous operation, you have to look beyond the robot to the bigger picture.