Technology Focus series: How will Driverless Cars Impact Rail in the Future?

Driverless cars are directed by an in-built computer, using artificial intelligence. These vehicles make journeys safer and ease congestion through smarter driving and vehicle to vehicle communication. They offer an opportunity to help rail overcome the ‘last mile problem’ but they must not be underestimated as an adversary.

Latest update: October 2019

Autonomous trains are now being trialled with promising results on heavy rail networks. Driverless cars are being permitted on roads around the world. Despite the overall positive outcomes there are some safety concerns due to accidents which may affect the future development of driverless vehicles. 

What are Driverless Cars? 

Driverless cars, or autonomous vehicles, navigate journeys without human input. The vehicle’s in-built AI (artificial intelligence) uses sensors to detect obstacles and road signs. The sensor system is often a combination of video cameras, Light Detection and Ranging (LiDAR) and radar. Using these, the AI is able to produce a full image of its surroundings and make appropriate decisions for the course of action required to travel from A to B safely. The AI is trained through machine learning to drive in the required manner and respond to dangers instantaneously by identifying threats.

The Society of Automotive Engineers (SAE) has produced a definitive hierarchy of autonomy. These are a set of guidelines that describe the levels of autonomy in driverless cars. As shown in the tale below, Level 1 is the most basic through to Level 5 which is the most advanced. 

Image author: Runner1928, via Wikimedia Commons.


What industries use this technology?

Cars with partial autonomy (up to level 3) are becoming increasingly popular for domestic use. Tesla cars have an ‘Autopilot’ feature, which includes a function whereby the car drives itself out of a parking space towards the user. Other autonomous road vehicles include driverless pods, which travel along dedicated guideways, an example is at Heathrow airport where they ferry passengers between the terminals and the car parks. In Greenwich, a similar technology is being trialled using cycle paths. The latter will explore how slow-moving autonomous vehicles cope with sharing a route with pedestrians.

Semi-autonomous lorry convoys, where only the lead lorry is driven by a person, have been given permission to use UK roads for trials. This method, referred to as ‘Platooning’, is expected to reduce congestion and emissions through more economical driving. However, there are concerns that platoons may cause issues for other road users by blocking exits and obscuring road signs. In Sweden, a driverless truck with level 4 autonomy is being trialled on public roads and is making daily deliveries. This offers the logistics industry a solution to the growing shortage of lorry drivers. 

Automated trains are common on metro systems. The Docklands Light Railway (DLR) in London operates at a level 3 Grade of Automation (GoA), with a train attendant on board to take control should any issues arise. On some London Underground lines, (including the Northern, Central and Jubilee lines), the driver is only required to operate the doors and initiate dispatch. Thameslink has also made use of this technology through central London, being the first mainline passenger service in the world to do so. In Australia, mining company Rio Tinto has successfully launched an autonomous freight train to transport payloads of 28,000 tonnes of ore across 280 km without human intervention. 

How will Driverless Cars impact the rail industry?

Autonomous vehicles can help in overcoming the ‘last mile problem’ to improve passenger services. They could collect and drop off passengers at the station as soon as the train arrives, making door-to-door journeys seamless. Such implications will make travelling by rail more accessible and increase rail use amongst passengers who may have previously encountered difficulty with rail travel, such as young families and disabled people.

Autonomous locomotives enable trains to run closer together because when such a train slows down it can communicate with the train behind it to reduce its speed in real-time. This will also allow for softer braking and acceleration, thus making energy use on trains more efficient. Such features have enabled, automated metros in Paris to run at an interval of only 85 seconds.

In the freight sector, developments could include driverless freight containers that load and unload themselves from trains could be produced. This would allow for the simultaneous unloading of several containers at once, which would make the process faster and more efficient.

What uncertainties remain?

In 2019 there was a high-profile case in which a Tesla driver was killed after his car failed to automatically stop when a trailer turned ahead of him.  An almost identical event had previously happened in 2016.  There was also an incident in Arizona where a pedestrian was fatally struck by an autonomous taxi. 

Concerns about how vulnerable driverless cars are to cyber-crime and security breaches have also been raised due to the potential for harm against both passengers and pedestrians.
Object perception is a limiting factor for full autonomy, as cameras can be sensitive to extreme sunlight. Defective traffic lights or infrastructure can disturb the on-board computer causing it to behave erratically in new situations, such as during power outages.

What is the current state of R&D?

Since each type of sensor has its own set of limitations, research is being conducted into combine the data these generate into a cohesive system of ‘sensor fusion’. The data can then be fed into the AI to anticipate how pedestrians and other road vehicles will behave. The University of Michigan has taught autonomous cars avoid collisions by using cameras to predict a pedestrian’s course of action based on their pose from up to 50 m away.

Driverless cars use neural networks to classify images and formulate an appropriate response. More reliable neural networks that can process data faster are being trained by the University of Glasgow.

What should the rail industry do?

Competition is the largest threat that driverless cars pose to the rail industry. The rail industry could attempt to combat this by focussing on delivering amenities that can be provided on a train but not in a car, such as warm food and toilet facilities, especially on long-distance journeys. Furthermore, the rail industry could focus on being the more convenient method of commuting to busy cities without the issue of congestion, and at being faster than a car over long distances. Finally, as autonomous rail vehicles become more prevalent and running costs decrease, services to rural communities, which tend to be less profitable, could be made more frequent.

Essentially, by integrating driverless cars into the rail business model could turn this threat to an opportunity. Not being limited to a specific location will enable driverless cars to collect and carry passengers from the train to their destination, as mentioned above in the Heathrow airport example.  

In Great Britain, the rail industry could develop long-distance autonomous rail routes which would allow the network to function more cohesively, with trains sharing live location and speed data, such as that being developed in Australia.

Log in or register to keep reading
Register for free individual access
  • Unlock research, articles and more
  • Get updates on RSSB’s activities





Need some help?
To talk to us about accessing RSSB content or corporate membership:
Cookies help us improve your website experience.
By using our website, you agree to our use of cookies.