Uber to deploy 500 sensor-equipped EVs for data collection

On Wednesday, Uber revealed a prototype car designed to gather real-world driving data for its growing roster of autonomous vehicle partners, which include Avride, Waymo, and WeRide. The company plans to deploy 500 of these sensor-equipped Hyundai electric vehicles globally this year. Uber expects 50 of these new data-collection vehicles to be on the road by summer. Once the fleet is deployed, it will be capable of collecting 2 million miles per month of high-fidelity data to help train self-driving systems.

The prototype vehicle is not a radical design, but rather a Hyundai Ioniq 5 fitted with sensors on its top and sides. These vehicles are retrofitted through a partnership with Roush Performance, which handles the physical modifications. The sensor suite fitted onto the vehicles includes:

• 14 cameras
• Eight solid-state lidar sensors—which use light pulses for remote sensing
• Nine radars

All of the gathered data will be routed through Nvidia’s Dual Drive Thor autonomous vehicle computer. Uber has indicated that it will continue to update this sensor suite as the needs of its partners evolve.

This initiative marks Uber’s first self-assembled vehicle project since 2020, when the company sold its former autonomous vehicle division to Aurora. The project is a key initiative for AV Labs, Uber’s internal division for autonomous vehicle data collection. AV Labs aims to develop a geographically diverse set of training data for autonomous driving. The company plans to share this high-fidelity data with its partners to improve autonomous driving software. To build this dataset, the division is analyzing data from fleet partners, including data collected from vehicles provided by Lucid in the U.S. and Europe over the past two years.

The data-gathering project is distinct from Uber’s other autonomous vehicle efforts. While AV Labs focuses on data collection to support its 30-and-counting autonomous vehicle technology partners, Uber launched a separate division in February called Uber Autonomous Solutions. That division is designed to handle the day-to-day operations of robotaxis, self-driving trucks, and delivery robots. Uber first disclosed its plans for the sensor-laden vehicle in January, prior to revealing the prototype on Wednesday.

Why it matters

Uber is positioning itself as the data backbone for the autonomous vehicle industry rather than just a ride-hailing platform. By leveraging its global scale, the company can provide geographically diverse training data to support its 30-and-counting partners.