Increased movement with maps of the future

hand and mobile phone with map of city

Most of us have digital maps at our fingertips today, on our smartphones. But these maps are rarely suitable for guidance on foot or for finding your way in new environments, for example as a tourist in a new big city. Using applied AI, researchers are now looking at developing maps that are adapted to encourage more movement.

One problem in big cities today is that there are too many vehicles, which creates both congestion and a poor environment. Various initiatives are trying to encourage more walking. In addition to improving pavements, footpaths and lighting, it has been recognised that the way information about navigating on foot is communicated is important so that residents see walking as an attractive option.
“We want to investigate how artificial intelligence can both give us researchers new knowledge and at the same time contribute to innovation and reduced production costs,” says Lars Harrie, Professor of Geographic Information Science, Lund University.

Tens of thousands of map examples needed

Understanding how maps are designed and how people search for information is a specialist skill as maps can contain many different layers of information.
“One of the challenges in today’s map production is that the placement of icons and texts and the like is handled manually and is very costly,” explains Lars Harrie.

In the project, Rachid Oucheikh, a postdoctoral researcher in machine learning, is working on various new methods for how a data-driven approach can create improvements and simplifications. It requires tens of thousands of maps to be read and analysed, with the aim of training an AI network to recognise good placements of icons and texts.
“We’ll approach the problem from multiple approaches. The one we are working on first involves transferring or translating information from an image source to a target image and doing so in a way that retains relevant content,” explains Rachid Oucheikh. “We will then compare this with another method, which involves identifying the main key positions on a map.”

We hope to contribute to new open tools that can be used by many

Lars Harrie

Test data from big cities

A big challenge in general is to get access to enough map data. Since maps are everywhere and digitisation has been going on for a long time, you’d think that wouldn’t be a problem. But most of the time, map data from public authorities and private companies have user licences that do not allow large-scale machine learning. Map data exists but it is usually not open or accessible.

“That’s why we are so happy about the cooperation with T-maps, which has many big cities as customers, and through them we can now test with real data in our project,” explains Lars Harrie.

The Swedish company T-kartor has its origins in Skåne but operates on a global market and makes tourist maps for major cities such as New York, London and Paris. The collaboration between the company and the researchers is valuable for both parties. T-kartor gains access to cartographic knowledge and methods that can lead to innovation of new advanced geographical products. Researchers gain a better understanding of the practical application and the different problems that need to be addressed.

“We collaborate with companies and across different disciplines in academia to both understand what the needs are and to develop new methods based on machine learning,” says Lars Harrie.

Tacit knowledge is heard

During the two years that Rachid Oucheikh will be working on the project, there will not be time to develop practical applications from the theories, but it is still of great value to be able to use modern technology to develop maps in the longer term. Only the City of London has a 100-page document of all the rules governing the design of maps, with instructions of various kinds based on numerous user tests. Often experienced cartographers have this kind of tacit knowledge intuitively but it is difficult to translate into formalised knowledge. Machine learning can provide answers to this by studying large sets of well-designed maps.

“It’s common to get what we call hidden knowledge using applied AI,” says Rachid Oucheikh, who has two previous postdoctoral positions behind him in Norway and in Jönköping, where he specialised in self-driving vehicles.

Information management in urban maps is a rather unexplored field in terms of machine learning application and the intention is to share results and insights as open source code for others to use and refine.

“Although we are collaborating with a company, they are not the only ones to benefit from our project. We hope to contribute to new open tools that can be used by many and that can reduce costs overall,” concludes Lars Harrie.

If better map information also leads to reduced environmental impact as more people choose to walk in big cities thanks to better guidance – that’s an added bonus to hope for.

Original text by: Marianne Loor