Pilot: Our team is using AI to plan the transportation network of the future

As Ray Sono, we work in the field of mobility for various customers and projects. We are constantly confronted with the topic of artificial intelligence (AI). Recently, however, we have moved into an innovative setting that combines mobility and artificial intelligence and is – both in terms of content and organisation – previously uncharted territory for us.

Katharina Große-Schwiep, Senior Consultant | Digital

Reading duration: 5 minutes

The Federal Ministry of Economics (BMWi) is currently organising a competition on artificial intelligence as a driver for ecosystems and is requesting applications for project funding. Nothing easier than that – we thought! We joined a consortium led by the Hamburg Hochbahn and are participating in this competition with the KIMOB project (“Artificial Intelligence Mobility”). After our discussions only went in circles, we threw our agile mindset into the Spree and set ourselves the goal of using AI to change Hamburg’s future traffic system. Getting an overview of complex systems, thinking from the user’s perspective, translating this into digital terms: This is our daily bread and butter. Now we’re at the starting line, we’re already enormously proud of what we have achieved, and want to continue. For three good reasons:

1. We want to get into other areas!

Everyone is talking about it, but no one is doing it. Changing a traffic system cannot be achieved by talking about it; it must be actively managed. Such control must take place at many different levels. The KIMOB project quickly showed that our ecosystem is complex, and that many different interests affect us. Of course, there is EU and federal policy, there are the Fridays for Future, and there is a social debate about particulate matter and nitrogen oxides. Ultimately, however, we must all get to work, take our kids to day-care, and go to the supermarket. No one can afford downtime.

In the first phase of the project, KIMOB will focus on the city of Hamburg. The city itself has a Senate Chancellery, an Environmental Agency, a Transport Authority, various mobility companies for bus, train, ferry, bike-/car-/scooter-sharing, taxi companies and delivery services. And many more. But of course, everyone who moves around in the city daily is especially important to us – in other words, road users. This includes drivers, public transportation users, cyclists, tourists, parcel delivery companies, and every worker with a van. What I’m saying is that the system is complex, its needs are diverse and sometimes in complete opposition to each other, time is short, and politics is under pressure. I affectionately call this “explosive tension”. In the project, we can resolve this tension via a user-centred approach and close cooperation with all parties involved. This is the only way to find solutions and build products that will actually be accepted and used, and which will contribute to our goal of bringing about a turnaround in Hamburg’s traffic.

2. We have the solution: Our artificially intelligent agents

I briefly enter the route in Google Maps and already know when I have to leave work to be on time for my dentist’s appointment. And I can see that I’ll be much faster on my bike than with the S-Bahn. What sounds easy to us as road users is a complex task for traffic planners because it is difficult to simulate the decisions of individual transport participants.

To coordinate bus lines, set up replacement rail service, or initiate a new underground line, traffic planners require an incredible number of data sources, which are in different qualities and conditions. In the future, we will use artificial intelligence to facilitate the work of traffic planners and improve their results. An AI-based agent system is being set up and trained so that simulations and forecasts can even be created in real time. In simple terms, an agent system maps traffic via autonomous software units. Each traffic user is represented as an agent who has goals and makes decisions based on their characteristics. For example, let’s simulate a big event like a Rolling Stones concert with several tens of thousands of people moving toward the city park within a very tight time window. We can see how the different agents adjust their behaviour. Experienced Hamburg residents avoid certain routes so they don’t get stuck in slow-moving crowds. Others change their modes of transport. At the same time, Hamburg visitors use the specially equipped shuttle buses, look for parking spaces with their own car, and in the worst case, look for the right S-Bahn whilst having absolutely no orientation. Perhaps Rolling Stones fans and thus their agents will have the special characteristic of arriving with their own drums, which in turn would have serious consequences on traffic. Everyone knows that such events put transport systems to hard tests. Therefore, having well-founded simulations in advance is especially valuable in order to initiate the right measures and optimally control the agents, i.e. the real people. With the agent system, we provide traffic planners with a tool that helps them make better decisions and optimise traffic planning by dynamically adapting their services. Individual road users also benefit from this, as their individual travel chain is adapted to suit their needs.

3. A Tinder match with experts

Everywhere we look: Experts. Experts in transport policy, for machine learning, for agent systems or data modelling. In particular, Hamburg universities (UHH, HAW, HCU) demonstrate in-depth research knowledge on artificial intelligence. At the same time, the regional company Geoinformation und Vermessung has been collecting great quantities of data on congestion and traffic density for years and making it available via a central data platform. Telefónica NEXT is preparing mobile data on anonymised movement streams to be used for planning public transport systems or infrastructure. Fantastic – so much domain knowledge in the consortium, such a crystal clear data pool and so many use cases to think about! And it is precisely these many different views and ideas that present us with a “positive dilemma” that needs to be resolved. All the experts must understand each other and work together to find out what most influences our goal – changing Hamburg’s traffic system – and what is technically possible. We want to develop a real solution and not reinvent Google Maps. That’s the real challenge. The basic idea of a consortium, in which various experts contribute their respective know-how, is enriching and exciting – and changes the way we all think. As Ray Sono, we are very close to university research and develop digital and needs-oriented products based on their results. At the same time, universities are moving into our world of real applications, where they can test their results in the existing world. It sounds like a Tinder match with the prospect of wedding! From now on, we’re keeping our fingers crossed! A few days ago, as part of the consortium led by the Hamburg Hochbahn, we submitted our competition entry. We are now waiting anxiously for feedback on whether we will receive funding and be allowed to implement it starting in 2020. We’ll keep you up to date!

KIMOB team

In Munich and Berlin, our KIMOB team worked on shaping the mobility of the future of another city with millions of inhabitants: Hamburg. At his desk in the office and live at the Hamburg location, project manager Norman Rockmann, tech consultant Dr. Max Franzke, and strategy consultant Katharina Große-Schwiep planned how traffic could soon flow through the city on the Alster whilst saving more resources through the help of artificial intelligence.

You want to learn more about Ray Sono? Get in touch!

You want to learn more about Ray Sono? Get in touch!

Nancy Forner
Marketing & Communications
Contact us here

Further stories