Autonomous trains: easier than in road traffic

Contents

Recently, we took a look in the rear-view mirror and considered over 20 years of development in autonomous cars. In doing so, we left out over ten years of previous research in the Prometheus project, in which full driving automation with camera systems was tested from 1994 onwards. This long period of time and today's reality, in which a human-operated cab is still much cheaper than an expensive robo-car, shows the difficulty of the task. In free road traffic, there are simply too many movement options. In rail transportation, on the other hand, the complexity can be reduced to a manageable level. This is why fully automated train lines have been in operation in Japan since the 1980s. A veritable global boom began around the turn of the millennium. Today, 47 cities worldwide operate autonomous train lines. In Germany, the first driverless subway train arrived in Nuremberg in 2008. Its example demonstrates several aspects that go far beyond "I can get rid of a paid person".

The U3 line in Nuremberg could only be created because it shares a section of track with the existing U2 line. If this had been solved with human driving, the frequency on the U2 would have had to fall in order to maintain the necessary safety distances. The planners wanted to avoid this. With this restriction, the only solution was to automate both lines. The implemented system clocks at 100  seconds per train. This results in double the throughput compared to human control. The driver's cab remains empty on these trains. In the event of exceptional disruptions, the human intervenes via the control center. These include medical emergencies, emergency stops and the manual unlocking of individual doors if the computer does not see a safe switching threshold, but the control center does. Travel, braking, points setting, signal switching and door control are automatic. Frankfurt is currently following this example, and Hamburg is planning autonomous local transport lines. In addition, the shunting yards where the vehicles are parked outside their times of use are currently being automated step by step.

Hidden advantages

We spoke to experts from the manufacturer of the system: Siemens Mobility. Siemens Mobility not only builds trains, but also develops control and safety technology for the automation and digitalization of rail infrastructure and vehicles. This has a whole range of advantages, not all of which are immediately obvious.

Reduced train intervals / higher frequency:
Operators can use existing infrastructure up to 30 percent better instead of having to build new tracks. In Copenhagen, for example, train intervals on the same line have been reduced from 120 to 90 seconds. It is also often difficult or even impossible to build new tracks. This applies to some areas in mountainous Switzerland, for example. In such cases, automation is the only option if throughput is to be increased.

Optimum efficiency:
Automated trains create significant energy savings in the region of 30 percent. They achieve this by adhering precisely to their timetables, accelerating and braking only as much as is necessary to fill the clock. People have to estimate. The computer has the real-time data. The system also brakes and accelerates the trains with an accuracy that a human being cannot achieve. It also reduces the load on the power lines: A starting train draws very high starting currents. Automation therefore combines the braking current fed back from one train and the acceleration of another train so that peak loads can be avoided in the majority of the system.

Smoother travel:
Smooth, precise acceleration and braking increase comfort and reduce noise, energy requirements and the mechanical load on all systems.

Flexibility:
The control system can automatically determine capacity utilization and use this to generate suggestions for statistically optimal timing. This could result in either optimized timetables or - thinking further into the future - a completely flexible system that breathes.

Lower costs:
All of the above points can reduce costs. Many people do not realize that rail infrastructure is quite expensive and therefore constantly supported by the state. So if better technology can reduce costs, it's better for all taxpayers, not just rail users. In order to achieve Germany's climate targets, such a thoroughly digitalized railroad also holds considerable potential, because rail transport is much more energy-efficient than road, water or air transport.

"The question is always: what is the alternative?" says Andre Rodenbeck, Head of the Infrastructure Division at Siemens Mobility. "Politicians set targets for the transport sector, and then the best solutions have to be found to achieve these targets. If you want to get more passengers on the rails, the cheapest and most effective way to achieve this is through more digitalization: automation technology, networking, real-time information, an overall simple and good rail transport experience for all passengers. This has been lacking so far. But there is a spirit of optimism and the signs are good. For the first time, for example, rail infrastructure is receiving more money in the federal budget than road infrastructure."

Near and far, 2 and 4

In the background discussion, Siemens Mobility divided the areas of application of train automation technology into local and long-distance transportation and into the train automation levels (Grade of Automation, GoA) 2 and 4, because these are the levels with the highest practical relevance at present. In GoA4, the train runs fully automatically and there is no need for a driver to be on board. In GoA2, software takes over the normal driving tasks from start to stop. The driver takes care of the doors, initiates the start and monitors the journey for unforeseen events.

Track sections in urban metro operations can often be secured so well that a very high level of safety can be achieved with comparatively little technical effort. There are therefore already some GoA4 systems in this environment, similar to the initial example in Nuremberg. In long-distance transport, however, the routes are too long and too complex to be constantly monitored everywhere. A system on GoA4 in long-distance traffic needs obstacle detection, for example. On high-speed roads, it would have to look very far ahead. "For Level 4, we have to replace the driver's eyes with technology," says Dr. Ireneus Suwalski, who is responsible for automation technology at Siemens Mobility. "Let's say a car gets stuck at a level crossing or another unforeseen situation suddenly arises. In Berlin, someone once threw a shopping cart onto the track from a bridge. You have to be able to react to that."

Anticipatory driving

A really good solution is not yet in sight, although the technology is constantly improving. "In a car, it's enough to monitor 200 m ahead," says Dr. Suwalski. "In a train, however, you need to be able to look much further ahead on fast routes in order to be able to initiate a meaningful reaction."

However, humans are also unable to recognize an obstacle kilometers ahead and brake to a halt in front of it. "Exactly," confirms Suwalski. "A human being may be able to detect a signal or obstacle at a greater distance than current sensors if visibility is good, but this is still the case in high-speed train operation: A train driver supported by automation technology still has the better options here to react and act. Even if a machine could do it better than a human in the future, and I believe that we will get there technically, there is still a question for society: do we want to allow machines to take on this responsibility?" For these reasons, GoA2 is currently establishing itself as a good solution in long-distance transportation: The software takes over the basic tasks and optimizes them as only a computer can. Humans take responsibility for safety and all work for which a higher-level context is relevant, which can start with door control.

More technology

Higher levels of automation mean a massive upgrade of the technology on the lines, in the control centers and in the railcars. The European Train Control System ETCS forms the basis for international interoperability. ETCS is installed in the control centers, but also as an additional, highly fail-safe computing unit in the railcars and is intended to replace the proliferation of previous train control systems in the long term. Communication will be via digital radio using the GSM offshoot GSM-R (R for "Rail"). ETCS is one of the cornerstones of Europe's rail infrastructure digitization and also provides a solid basis for further automation. This "Automated Train Operation" (ATO) takes place in additional, separate computers for reasons of safety and redundancy.

Dr. Suwalski explains: "An automated system on GoA2 relieves the driver through automation ETCS monitors the journey and thus further increases the safety of the system." Similar to the ETCS equipment, the additional ATO systems only consist of computers and communication technology, in this case mostly a 4G radio network. Additional sensors are only used from GoA3 onwards. At this level, the human only works as a supervisor and the system should cover the majority of tasks on its own. However, with the current state of technology, this still requires fully controllable routes. Free long-distance transport will therefore continue to require people in the driver's cab for a long time to come.

Outlook

The current upgrades to ETCS, automation to GoA2 in long-distance transport and to GoA4 on appropriately safe local transport routes and the associated digital infrastructure are the major current developments in rail transport automation. A fully automated long-distance train (GoA4) currently offers fewer advantages in this environment than one might think. The cost structures are particularly difficult. Seamless route monitoring would be very expensive. But it is also about the number of trains. The total global market for railcars is in the region of 3000 to 4000 units per year, of which only 50 are high-speed railcars. With such quantities, hardly any economies of scale can be expected, such as those achieved by car technology. In addition, a train can easily see 30 years of daily operation. However, you can turn this around and see the circumstances as an advantage: Every new train system can (indeed: must) be designed to suit the transportation task to be solved. And here we will see some progress in the coming years - most of it through unrivalled cost-efficient digitalization measures.

(cgl)