Are the wheels in order? Do the brakes work? Does the side wall have a dent? The DB wagon inspectors are on the move at all times. With the wheeltapper’s hammer they check the wheels for damage, take a close look at all the wagons before the departure of a train: on warm days, in cold nights, in the rain, the snow and storm. This does not look like an ideal place to employ digital innovations. But it is precisely in this area that DB Systel, in close collaboration with DB Cargo, has launched one of its projects for the promotion of paperless transport: “Dialogue on track” is developing a language assistant for the business that can be used in many ways.
The idea behind “Dialogue on track” was developed over the course of the train formation facility 4.0 project (ZBA 4.0), a comprehensive digitalisation campaign by DB Cargo. ZBA 4.0 not only comprises major undertakings such as the automation of shunting locomotives, coupling robots and video analysis of passing freight wagons, but also projects that make the everyday working life of employees more effective and pleasant – like the products introduced by “Dialogue on track”.
“We want allow them to use both hands”
In daily rail operations the wagon inspector plays a key role. No train is allowed to leave without their permission. “A wagon inspector has to handle a variety of tools each day,” said Stephanie Blum, product owner of “Dialogue on track”. Along with all the tools they have to carry with them, the wagon technician also uses pen and paper for handwritten notes as well as tablets as of late, to document the damages. They put their gloves on, then take them off, take out their tools and put them away. The work process is marked by continuous interruption. And then the railway worker has to go back to the office as well. This is not the way to get into a flow state.
“There was an idea to reduce these media disruptions,” said Stephanie Blum. “Would it be possible use language-assistance on this process?” In light of digital language assistants becoming ever more popular – just think of Siri, Cortana or Alexa – a dialogue-based language input and output device for the wagon inspector appeared to be the right way to help with the recording of all process-relevant information. “We want them to be able to use both hands,” said Sven Seligmann, head of Working Procedures Ground Staff at DB Cargo. Interruptions would be done away with and back office work would also be minimised. The downstream production systems would receive all the necessary information via the interfaces and would do so in real time. The wagon inspector would find work easier, it would be quicker and continue to be safe. So this is an ideal case-study of process optimisation.
Test, test and test again
The idea was first successfully pitched at DB Systel and “Dialogue on track” was initiated in December 2017. Employees interested in the project got together, including an expert on user experience. “After all, we all know that our goals can only be achieved if the user is taken on board,” said Stephanie Blum.
DB Cargo supported the “Dialogue on track” team from the start. Submitting “Dialogue on track” to the DB Systel Skydeck Accelerator programme secured further support in fleshing out the business cases as well as refining the prototype. “We found a key user for “Dialogue on track”, who agreed to test the first prototype of the language assistant,” said Frank Weppner, head of Asset Automation in the ZBA 4.0. programme at DB Cargo.
In a user journey with a former wagon inspector turned team leader (the key user), the process was meticulously analysed, so that the requirements of the prototype became quite clear. Furthermore, a field test in Mainz-Bischofsheim was carried out to analyse the influence of background noise on the use of a language assistant. The wagon inspectors on site were very impressed by the idea that a language assistant could support them in their daily work. Another field test was organised by DB Cargo in Seelze near Hanover with three wagon inspectors, in order to once again test the prototype. “The employees were impressed. Of course, not everything went 100%. When one of the wagon inspectors tried a Saxon dialect on the system, it went mute,” said Stephanie Blum. But the testers’ response was positive nevertheless.
Brought to you by AI
A central role in the development of Dialogue on track is played by artificial intelligence (AI). “The language assistants we know from applications at home, such as Alexa, Siri and others, would not be able to cope. We need something that can learn the language of the wagon inspectors,” explained Stephanie Blum. “They should be able to speak as they wish – not as the device wishes. AI helps the system to understand what the user wants. With thousands of language entries per day the artificial intelligence learns to suggest the right things, which the wagon inspector can then confirm,” said Sven Seligmann.
Without AI, employees would have to be asked how they would describe a certain type of damage, for example, and then completely update the system with these sentences – a substantial manual effort and a major source of errors. “The priority is to ensure that what the device does meets our requirements,” said Sven Seligmann. “Along with process improvement and a reduction of workload, safety is most important. In the worst case a wagon might set off that should have been withdrawn from service. Such risks must be eliminated during the development phase of the device and not be amended afterwards,” added Seligmann.
The users have the last word
The language recognition of the system is now being tested in five different regions. “If everything goes smoothly, we will evaluate further application cases and take a look at possible end devices for operation,” said Stephanie Blum, outlining the further steps. DB Cargo is also interested getting other business units on board to make the project scalable and feasible. “After all, the digital innovation we are developed for the wagon inspectors also applies to other areas at DB. A solution that only works for a quarter of all employees does not really get us anywhere,” explained Frank Weppner.
When it comes to drawing in more partners, the participants are optimistic. “We are developing a product that can be used in many ways,” said Stephanie Blum. “It can be used wherever an employee needs their hands free to check, look at or touch something and needs to document the results. Such processes are not limited to DB Cargo. Long Distance, Regio or the depots, among others, have them as well.”
No decision has been made as yet as to what the “Dialogue on track” language assistant for the DB Cargo application will ultimately look like. A “voice-only” control solution seems rather unlikely. According to the project participants an additional screen interaction is more likely. This would give the users more opportunities for corrections. “Of course, the device should optimise processes and further increase quality,” said Sven Seligmann. “But the user always has the last word. That’s important.”