Uber is accelerating the development of its autonomous driving data infrastructure. The company unveiled a prototype vehicle for road data collection and plans to deploy 500 modified Hyundai Ioniq 5s globally this year to serve its expanding network of autonomous driving partners.
These vehicles will collect real-world road data for partners such as Avride, Waymo, and WeRide. Uber says it expects to have 50 vehicles on the road by this summer; once the entire fleet is built, it will be able to collect approximately 2 million miles of high-precision data per month.
The data collection vehicle was assembled under the leadership of Uber.
Judging from its vehicle design, this is not a completely new model, but rather a Hyundai Ioniq 5 with numerous sensors added. Multiple sets of equipment are located on the roof and sides of the vehicle to record more complete road environment information.
Uber stated that this is the first time the company has assembled such vehicles itself since selling its self-driving division to Aurora in 2020, with the assembly process assisted by a partner. This also signifies that its newly established AV Labs division is entering a more concrete execution phase.
Sensor and computing power configurations announced
According to Uber, these vehicles will be modified through a partnership with Roush Performance, which will be responsible for adding hardware to the vehicles.
- 14 cameras
- 8 solid-state lidars
- 9 radars
The data collected onboard will be integrated into Nvidia's Dual Drive Thor autonomous driving computing platform. Uber also stated that it will continue to adjust the sensor mix based on partner needs, rather than sticking to the current version.
The goal is to expand training data coverage.
Uber says its goal is to build a more geographically comprehensive autonomous driving training dataset and provide partners with time-synchronized, fully stitched road views for training autonomous driving software.
The company also revealed that it has already accumulated a certain foundation. Previously, Uber had collected data in dozens of cities through thousands of dedicated vehicles; in the past two years, the company has also obtained relevant data from hundreds of Lucid Air vehicles operating in the United States and Europe.
Currently, AV Labs is analyzing these two sets of data and plans to further expand the data collection scale by retrofitting a new batch of Ioniq 5 robots. For Uber, this strategy is not about directly developing its own Robotaxi, but rather about extending its platform capabilities to the autonomous driving data supply chain.
