Facebook Extended Its Map With AI Tool Suite to OpenStreetMap

The social network is using artificial intelligence to predict road networks

Facebook AI researchers and engineers were able to add over 300,000 miles of missing roads in Thailand
Facebook

Facebook has been using artificial intelligence to predict road networks and create maps via commercially available high-resolution satellite imagery, and the social network is making its model available more widely.

The company said in a blog post that its Map With AI set of specialized map editing services and tools—including RapiD, which enables mapping experts to review, verify and adjust maps created via its findings—is now available to community-based effort OpenStreetMap.

Research scientist Saikat Basu; software engineers Derrick Bonafilia, James Gill and David Yang; and engineering manager Danil Kirsanov said in the blog post that Facebook AI researchers and engineers were able to add over 300,000 miles of missing roads in Thailand and more than 90% of missing roads in Indonesia to OSM, completing the project in 18 months.

They described OSM as the largest free and open map data source in the world, saying that thousands of volunteers catalog missing roads, buildings and bridges, and cautioning, “Millions of miles of roads around the globe have yet to be mapped. Map data gaps can negatively impact everything, including disaster response, community planning and helping the local economy. It’s been a long and manual process until now.”

Users of the RapiD tool can select a road to be added to the map, at which point it can be further edited before being submitted to OSM. White lines represent existing OSM roads, while magenta lines represent predictions.

Facebook said in its blog post that accurate mapping data will also help it better serve people with its products, including Facebook Marketplace and Facebook Local.

The social network added that, as is the case with the population density maps project it unveiled in May, maps created via this initiative will be publicly available as a resource for disaster response, urban planning, development project and other use cases.

For much more on Map With AI, RapiD and OSM, see the blog post by Basu, Bonafilia, Gill, Yang and Kirsanov.

Recommended articles