Automatic for the people? ATO’s positives and pitfalls

Thursday February 16 was a day of irony. ASLEF revealed that drivers working for Southern had rejected the deal the union had presented them. Meanwhile, in the Institution of Mechanical Engineers in London there was a conference. It’s subject? Automatic Train Operation (ATO) and driverless trains.

I suspect they’ll not be here anytime soon, except on Docklands Light Railway, but ATO is coming to Thameslink and it’s been on London Underground since 1969. Nevertheless, it’s a short step from having a computer driving a train with a driver keeping an eye on things to not having a driver at all.

Newspapers and television shows have showcased driverless cars for a while. IMechE conference delegates were treated to a very stylish film from Mercedes Benz of a driverless truck, courtesy of a DB Cargo speaker. This truck was seen rolling along an autobahn (presumably a test track given the lack of other traffic) with the driver happily checking an electronic tablet.

The DB speaker followed it with another clip, this time of a slightly snowy Siemens test track at Wildenrath of tests of a driverless freight locomotive last March. It gently buffered against some wagons, using LIDAR detection to measure distance. It slowed for a speed restriction and it stopped short of an obstacle in the track, courtesy of on-board radar. Of course, it still needed a human to throw the shackle over the wagon’s drawhook and connect the air brake pipes but that wasn’t the technology being tested.

This being railway technology, there’s another load of abbreviations to master. In this case, it’s ‘GoA’ or ‘Grades of Automation’. They run from one to four. GoA1 describes Great Western’s inter-city trains, a driver is in full control but automatic train protection (ATP) prevents him passing red signals. GoA2 adds ATO so the train stops and starts with the driver opening and closing the doors and taking over if there’s a problem. GoA3 has no driver but a on-board attendant can take over if there’s a problem. This is equivalent to Docklands. Finally, GoA4 has no crew at all, like some airport shuttles.

Britain has added an intermediate stage, GoA1+, which means there’s a driver advisory system given the driver information about the best speed to drive to match the timetable.

ATO is common on metro systems. LU’s Victoria, Central, Jubilee and Northern lines use it with a driver present. Shifting it to main line services might appear simple but in practice it is not. ATO can be done on metro routes because they are generally simple and generally have one type of train. The Victoria Line, for example, will only have 09 Stock running. Main line railways are more complicated with many types of stock permitted to run.

Thameslink represents a half-way house. It will use ATO in its core section between St Pancras and Blackfriars. Only one fleet will run in this section. Thus it has the characteristics of a metro system but on a mainline railway. It’s using ATO because NR decided that was the only way it could achieve a consistent throughput of 24 trains per hour. ATO provides consistent performance in place of each driver being slightly different. It can be programmed to drive on the limit of a train’s performance, accelerating for as long as possible and braking hard as late as possible to get the most from trains and tracks.

Or it can be programmed to drive smoothly, reducing wear and tear, but still meeting the timetable. It doesn’t need a gap in a timetable every hour to ensure reliability because ATO trains run consistently. Trains using Thameslink will have complete details of their network loaded into their computers. Distances, gradients, station locations – all that sort of thing. But try loading the UK rail network into the computer on board a Class 66.

Try programming into a computer all the possible permutations of trains approaching junctions at the same time. Which should go first? Perhaps the one that gets there first? Or the one that’s on-time. But what if the other is late but has more people on board that will be delayed? What if the first one is a ‘stopper’ but the second an express?

These are decisions that signallers and controllers make every day. They use guidance and experience. An ATO system can only rely on what its programmers tell it. If Britain is to implement ATO across the country, someone will have to work all this out.

Will passenger trust ATO? They appear to trust airliners that can take off and land under computer control. They appear to trust Heathrow’s computerised air traffic control. We get a bit wobbly when it comes to autonomous cars and I’ve no idea how we feel about a 44-ton truck bowling down the road as its driver surfs the net. But I might guess.

We happily clamber on board LU’s automatic trains but I’ve a feeling we don’t know they’re automatic. DLR is pretty busy but it’s a small, slow railway. It’s also completely segregated with robust fences so the chance of trespass by people or animals is low. Track access by staff will be strictly controlled.

But what if the Scotch Express or the Cornish Riviera was run by a computer? It’s not likely to come off the track as an autonomous car might leave the tarmac and plough over a pavement, perhaps to avoid a child running into the road. That’s where that swish and comforting Mercedes Benz film comes in. It exudes confidence and control. It reassures.

Where is the railway’s equivalent? A train gliding through countryside. Passengers sitting in comfort. A perfect stop at every station. And all under the unwavering eye of a computer.

It’s only with this reassurance that passengers could be convinced to accept driverless trains. While the engineers keep working to master the technical challenges perhaps the wider railway can start to explain what the future of travel might look like.

This article first appeared in RAIL 821, published on March 1 2017.