There is a race to develop autonomous cars. These cars can run only on the basis of guidance from the machine-readable maps known as splines or digital rails. The market for maps for autonomous cars will grow in value from around 92.2 billion in 2020 to $ 24.5 billion by 2050. As Google is a dominant player in consumer mapping, it has a competitive advantage in this emerging field. It has an autobnomous car-spin-off called Waymo.

There are three key imput in digital mapping. First is the information about roads and buildings. These are commoditised maps. In such maps OpenStreetMap which is a British open data repository and OpenAddresses, its cousin, are widely used. Many mapping busineses source the data from OSM. Second key input is the close up detail of the streets. Here a machine learning technology called deep learning is used to scan the imagery — as many as 80 billion photos. This identifies house numbers and names of streets. Such imagery is being captured since 2007. The photos generated are in Streetview. Google and Mpillary, a Swedish start up collect street photos. Besides the autonomous cars will use laser scanners and radar to add to these images. Third, smart phone users using Google map as they move around provide large quantities of real-time GPS location data.