Have you already searched a base map of some part of the world and didn’t find it? It’s not surprising. Nowadays, many places in the world are still unmapped and, trying to create all these missing maps is a time-consuming process even with good satellite imagery and good mapping software. To accelerate the production of those documents, Facebook decided to work with one of the best free base map producer OpenStreetMap. Together, they use artificial intelligence with high-resolution satellite imagery to draw roads across the world.
Map With AI
An Artificial Intelligence or AI is a computer capable to reproduce tasks that normally require human intelligence. In the mapping world, it means that computer and software are able to create maps without human intervention.
Facebook AI researchers and engineers developed a collection of map-editing services and tools which are based on Artificial Intelligence. All the data produced are available to the mapping community through Map with AI. However, there are limits to what a computer can do, therefore Facebook decided to create a mapping tool for the OSM community named RapiD to map roads.
RapiD : Map faster, better and together

Picture 1 : Predicted Roads_Source : https://mapwith.ai/#13/8.6519/39.75361/0/55
RapiD is an extension for mapping on OpenStreetMap. It was built to be manipulated by different kind of cartographers, from the beginner to the professional. The first step doesn’t need human intervention, an AI analyses satellite imagery and predict the location of roads (shown in magenta on picture 1). After this step, the help of the cartographers’ community is required.
With RapID the cartographer doesn’t need to draw or modify the road if the IA predicted it well but, he has to select it and give some information about it. He must give the name of the roads and different elements shown in the next picture. Moreover, RapID allows to add some elements that the AI cannot draw. For example, you can pint points like a library, parks or shops (picture 2). The main goal is to create the most detailed map.

Picture 2 : RapID’s User Interface
RapID a tool to help after a disaster: Keala, India
Each year, from June to September it’s the monsoon season in India. During this period, rain is extremely important and sometimes creates disaster. In August of 2018, Kerala an Indian city was underwater. Severe flooding happened and affected millions of people. Unfortunately, base maps available for this part of the country didn’t identify every road of the area at the time. This lack of information gave a harder work the rescue team. So, to help the local OSM community ask Facebook to help them for mapping missing roads. Within a few days, 34 mappers from the Facebook will import over 21,500 roads in the Keala Map.
Do you want to help us?
If my article makes you realize that you can improve the everyday life of some citizen without leaving your couch. Join the OSM community and click here: https://mapwith.ai/#13/8.6519/39.75361/0/55 to draw some roads!
Bibliography
BASU.S, BONAFILIA.D, et al (2019) « Mapping roads through deep learning and weakly supervised training » Facebook Artificial Intelligence [Online] https://ai.facebook.com/blog/mapping-roads-through-deep-learning-and-weakly-supervised-training/
GitHub « RapID – an enhanced version of iD for mapping with AI » https://github.com/facebookincubator/RapiD
mapwith.ai « Using AI to Map the World » https://mapwith.ai/#13/8.6519/39.75361/0/55