Artificial intelligence is a hot topic not just at Deutsche Bahn. Anyone who visited the Mobile World Congress in Barcelona in late February will know that numerous technology vendors are now spicing up their products with AI. But that doesn’t mean the topic is new. The conference at which the term “artificial intelligence” was first used was held at Dartmouth College in England back in 1956. The victory of chess computer IBM Deep Blue over world champion Garry Kasparov in 1996 can be seen as the first high-profile appearance of AI in a real-world setting. In 2011, IBM Watson defeated two human champions in a quiz show, and in 2016, AlphaGo beat champion Lee Sedol at the complex game of Go.
But we no longer have to look to the media to find striking examples of AI in action. These days, we all carry artificial intelligence around with us on our smartphones – in the form of Siri, Alexa or Google Now. Today, the technology is so advanced that artificial intelligence is playing an ever-greater role in products and services.
Analysing and interpreting big data
This trend is being fuelled by high-powered computers and affordable cloud services. At DB Systel, artificial intelligence is also being deployed in many places. The latest trend tracker sees excellent prospects for the use of the technology, with some 40 projects currently under way in the AI space. And the involvement of AI will increasingly become the norm when it comes to analysing and interpreting large volumes of heterogeneous data and continuously optimising associated processes. “Many AI approaches can now be found in the big data ecosystem. This is because many of our use cases revolve around enhancing data analysis,” says Oliver Petrich from IT Strategy & Digital Transformation, who handles AI at Group level. He sees identifying and promoting appropriate potential, prototypes and applications within the Group as key.
Many AI approaches can now be found in the big data ecosystem. This is because many of our use cases revolve around enhancing data analysis.
Integrating artificial intelligence involves two distinct levels. On the one hand, AI can optimise and enhance existing use cases. If we consider data analysis, for example – in other words, data intelligence and advanced analytics – deploying AI can further enhance the efficiency of some solutions and processes. On the other hand, the use of AI in this area also allows totally new use cases to be discovered and may even enable totally new business models.
From speech to image
A look at the Periodic Table of AI can help to identify where deploying artificial intelligence makes sense. According to this source, various technologies are used in one phase to capture information – whether that be voice, audio, faces or images. In other words, this phase generally involves recognising things. The information is then processed and conclusions are drawn from it. Finally, these conclusions are output, or actions are triggered, either by voice or in other forms. The process is similar to the way people do things: we take information on board, process it and then state our conclusion. “This is why, when we talk about AI, we’re not talking about a single technology,” explain Sandra Buchheister, portfolio manager for artificial intelligence at DB Systel. And she adds: “The various technologies are at different stages in their development.”
For example, natural language processing – a technology used, for instance, in voice-activated assistants, voice control, dialogue systems or for analysing text – is relatively advanced. In this area, Deutsche Bahn not only has use cases and developments, but also its first standard products – such as DB Early Bird, an application that leverages information from social media and other publicly available data sources to identify and locate delays, traffic jams and flight cancellations. The Assistify chat system also employs AI for certain tasks: for example, to assist customer support or communication within teams. At the same time, there are major topics that require detailed consideration in order to make the most of potential in the areas of dispatching, timetable optimisation and autonomous driving. “Without question, these hold the greatest potential for the Group.”
Sharing knowledge and offering solutions
The market environment shows that DB Systel is on a par with other providers in many fields of AI. For example, online retailer Amazon uses standard and infra-red cameras in conjunction with AI in its AmazonFresh food delivery service to automatically recognise how ripe avocados are. “We’re pretty close to that,” says Oliver Petrich. Together with surveillance cameras, AI is already helping to detect heavy snowfall on platforms in winter and commission the snow-clearing service to deal with it. The technology also automatically anonymises individuals captured by the cameras. Video analysis using self-programmed algorithms is already delivering high recognition rates based on very little training data. “We may not analyse avocados, but we can recognise whether there’s snow on the platforms. And that’s certainly comparable in terms of complexity.”
We know the pros and cons of each particular technology, and this enables us to decide with the customer which technology is best for the specific use case.
First and foremost, these solutions can be used to come up with recommendations for other use cases. All business units are involved in this process: “What is important is that, within the Group, we all head in the same direction, share knowledge and offer the right solutions,” says Oliver Petrich. To achieve this, the DB Systel experts analyse which technologies are available on the market. “We know the pros and cons of each particular technology, and this enables us to decide with the customer which technology is best for the specific use case,” says Sandra Buchheister. The various individual topics should not be seen in isolation. Instead, they can be combined, as the example of the information robot FRAnny shows.
A machine that can communicate in multiple languages and responds to voice and gestures is clearly one of the most high-profile examples of artificial intelligence in action. But, as the examples show, AI is already being used for many other applications at Deutsche Bahn, often without this being immediately apparent. And this technological journey is only just beginning. Thanks to open source technologies and a strong community, develop made-to-measure solutions for the Group is becoming increasingly easy and affordable. One thing is already evident: the number of applications without AI will steadily decrease in the future.
Other articles on the featured topic of artificial intelligence: