Uber aims to become a platform that supplies data to autonomous driving companies by fitting sensors to vehicles driven by its regular drivers, TechCrunch reported on May 2.

Praveen Neppalli Naga (프라빈 네팔리 나가), Uber’s chief technology officer, unveiled the plan at an event in San Francisco. He said it was a natural extension of the AV Labs programme announced in January, the report said.

AV Labs currently operates only a small number of sensor-equipped vehicles that Uber runs directly, separate from vehicles driven by regular drivers. But the goal is much bigger. Uber has millions of drivers around the world. If even some of those vehicles can be converted into a data-collection platform, it can provide data at a scale no autonomous driving company can build on its own.

Naga said the bottleneck in autonomous driving development is no longer technology but data. He said companies such as Waymo need to drive cars themselves to collect data across scenarios, but lack the capital to do so. He said Uber can provide the data autonomous driving companies want under the conditions they want, such as data for specific intersections or time periods.

Uber has partnerships with 25 autonomous driving companies, including London-based Wayve. It is also building AV Cloud, a labelled sensor data library that partners can search and use for model training. Partners can also run Shadow Mode tests to simulate their models’ performance based on Uber’s real-world trip data, without deploying autonomous vehicles on public roads.