Uber wants to turn millions of its drivers into the sensor grid of self-driving companies

Uber has a long-term ambition that goes beyond transporting passengers: the company eventually wants to outfit human driver cars with sensors to get real-world data for autonomous vehicle (AV) companies — and other companies that can train AI models in real-world situations.

Praveen Neppalli Naga, Uber’s chief technology officer, revealed the program in an interview at TechCrunch’s StrictlyVC event in San Francisco on Thursday night, describing it as a natural extension of a startup program the company announced in late January called AV Labs.

“That is the way we want to go in the end,” said Naga as he equipped the drivers’ cars. “But first we need to understand the sensor kits and how they all work. There are certain regulations – we have to make sure that all districts have [clarity on] what senses mean, and what sharing means.”

Currently, AV Labs relies on a small, dedicated fleet of Uber-powered autonomous vehicles, separate from the driver network. But the desire is clearly too great. Uber has millions of drivers around the world, and if even a fraction of those vehicles could be turned into data collection platforms, the scale of what Uber could offer the AV industry would dwarf what any AV company could put together on its own.

The understanding that drives the system, Naga said, is that the limiting factor in AV development is no longer the underlying technology. “The bottleneck is data,” he said. “[Companies like Waymo] you need to go and collect data, collect different situations. You might know: in San Francisco, ‘At this school intersection, I want some data at this time of day to train my models.’ The problem for all these companies is to get that data, because they don’t have the money to install vehicles and go to collect all this information.”

Being the data layer of the entire AV ecosystem is a pretty smart play, especially when you consider that years ago Uber abandoned its ambitions to build self-driving cars (a move that its founder Travis Kalanick publicly lamented as a huge mistake). Indeed, many industry observers have wondered whether, without its self-driving cars, Uber may one day be rendered obsolete as AVs proliferate around the world.

The company currently has partnerships with 25 AV companies – including Wayve, which operates in London – and is building what Naga described as an “AV cloud”: a library of labeled sensor data that partner companies can query and use to train their models. Partners, which Uber plans to invest directly in, can use the system to run their trained models in “shadow mode” against real Uber trips, simulating how an AV would perform without actually putting one on the road.

“Our goal is not to monetize this data,” Naga said. “We want to do democracy.”

Given the obvious commercial value of what Uber is building, that freeze may not last long. The company has already made equity investments in several AV players, and its ability to provide proprietary training data at scale could give it significant leverage in a sector that now relies on Uber’s marketplace to reach customers.