French national railway operator SNCF aims to have driverless high-speed trains running by 2023.
The company is working on a ‘train drone’ project, and is hoping to have a prototype train to test the transport of goods between Paris and the south east of France in 2019, according to a report from FranceTV.
The driverless train will look like a regular train, but will be equipped with external sensors that can anticipate the slightest obstacle on the track, and if necessary, automatically brake. However, it will likely be semi-autonomous – at first at least – with a train driver in the control station, to take control in the event of unforeseeable issues and emergency situations.
Having a driver or someone on board to look after the train at certain situations is not new: in London, the Docklands Light Railway (DLR) has run like this for years. However, Matthie Chabanel, of SNCF, suggested that it would be a ‘world’s first’, perhaps because it is a high-speed train that travels at almost 200mph, and because it is aiming to become fully autonomous in the years to come.
Chabanel likened the system SNCF that is working on to a plane’s auto-pilot system.
“On high-speed, we are aiming for automation in the sense of automatic steering as in an aircraft. In aircraft, you always have a driver, fortunately, but you also have an automatic steering system,” he said.
The need for speed
Rather than reducing headcount, the aim, according to SNCF, is to increase the speed and frequency of high-speed journeys. It believes that the automated system would increase the number of trips between Paris and Lyon, for example, by 25 percent.
Rob Bamforth, analyst at Quocirca, believes that while autonomous trains are technically feasible, the biggest hurdle is passenger acceptance.
“Is there a risk and even if this is only a perceived risk, are we willing to take it?” he asks.
“The greater the numbers of people travelling in one ‘vehicle’ and the greater the speeds, the greater the fear and risk of a bad accident,” he says.
However, Bamforth suggested that is the number of variables can be reduced, then the simpler it should be to keep these trains both autonomous and safe.
“The more things are monitored and checked [to ensure] that they fit within the expected norm, the more processes are confirmed with explicit actions, and the more the risk is reduced,” he states.
However, he doesn’t believe that computers will be necessary for every aspect of autonomous trains; engineers will be needed to decide how many parameters need to be used and checked to make the vehicle autonomous, as developing the right systems that take into account feedback loops, redundancies and fail-safes involve substantial design skills.
“The engineering is evolving to make it feasible; the psychology and marketing needs to be there too, in order to make it acceptable,” he concluded.