San Francisco, CA — Uber's chief technology officer revealed Thursday night that the company aims to transform its millions of drivers into a real-time sensor grid, harvesting data to train self-driving systems.
Praveen Neppalli Naga announced the strategy during an on-stage interview at TechCrunch's StrictlyVC event in San Francisco, calling it a "natural extension" of the company's existing AV Labs program.
The plan leverages Uber's vast fleet of human-driven vehicles as mobile data collectors—a move that could give the ride-hailing giant a massive edge in the race to deploy autonomous vehicles.
Background
Uber first hinted at this direction in late January when it launched AV Labs, a new internal division focused on accelerating autonomous vehicle development.
AV Labs was presented as a hub for research and pilot programs, but Neppalli Naga's comments at StrictlyVC clarified the unit's primary mission: harness the data generated by Uber's own fleet.
Every Uber driver already carries a smartphone equipped with GPS, cameras, and sensors. The company now wants to tap into that hardware to collect road conditions, traffic patterns, and driving behaviors—all without requiring expensive retrofits or dedicated test vehicles.
Read what this means for the industry
What This Means
If successful, Uber's sensor grid could provide a continuous stream of real-world driving data far larger than any single autonomous vehicle fleet currently generates.
Rivals like Waymo and Cruise rely on a handful of purpose-built robotaxis operating in limited geographies. Uber's approach would blanket entire cities with data-collection vehicles.
"We're already collecting anonymized telemetry from every trip," Neppalli Naga said at the event. "Extending that to capture visual and lidar-level data from driver phones is a logical next step."
The move raises privacy concerns. Uber has stated that all data will be anonymized and aggregated, but critics worry about the boundary between driver monitoring and surveillance.
Neppalli Naga addressed these concerns: "Drivers will opt in, and we'll compensate them. This isn't about watching people—it's about understanding streets."
For the autonomous vehicle industry, Uber's plan could dramatically lower the cost and timeline for developing safe self-driving systems. Instead of waiting for dozens of test cars to clock millions of miles, Uber's network could deliver billions of miles of data per year.
The company has not specified a start date for the full rollout, but sources indicate that pilot programs are already underway in select U.S. cities.
Industry analysts say the move repositions Uber as a data infrastructure company, not just a ride-hailing service. If the sensor grid works, Uber could license the data to other self-driving startups or even sell access to its network.
Neppalli Naga declined to comment on monetization, but noted: "We're building the most detailed map of the world's roads. That has value beyond Uber."
The announcement comes as the autonomous vehicle sector faces a funding slowdown and increasing scrutiny over safety. Uber's sensor grid could provide the breakthrough needed to keep the industry moving forward.