RoadRunner in New York City.
What if there was a different way to do it?
What if we could build a process where you press a button and make a map instead? This process could take in huge amounts of the latest data and process it into a map. Then tomorrow, it could do it again. By throwing away yesterday's automatically created map, we no longer need the vast maintenance infrastructure and can just create a fresh new map every day. We would sidestep a large amount of conflation problems (merging maps) if we were clever.
The data that makes the map could come from GPS, cameras on cars, other maps, satellite imagery and so on. In fact, these could all be variables that go into map creation itself: I might want to push a button and make a map using satellite and cameras. You might want to press a button and make one using GPS and other maps.
This is exactly what we’ve done with RoadRunner, an early look at what automatic map making could look like.