Uber is setting its sights on a future that extends beyond transporting passengers; the company aims to equip its drivers’ vehicles with sensors to gather vital real-world data for autonomous vehicle (AV) developers and other businesses focused on training artificial intelligence (AI) models. Praveen Neppalli Naga, Uber’s Chief Technology Officer, unveiled this strategy during an interview at TechCrunch’s StrictlyVC event in San Francisco, highlighting it as a natural evolution of a new initiative named AV Labs, launched earlier this year.
Naga expressed that the long-term vision includes outfitting cars driven by humans with sensor technology, although he acknowledged the need for thorough understanding of how these sensor kits operate. Additionally, he mentioned the necessity of navigating regulatory landscapes to ensure that every state is clear on the implications of data sharing. Presently, AV Labs employs a small fleet of sensor-equipped vehicles managed independently by Uber, distinct from its vast network of drivers. However, the potential for expansion is considerable, as Uber has millions of drivers worldwide. If even a small percentage of these vehicles could be converted into data collection units, the massive volume of data collected could significantly surpass what any single AV company could independently gather.
The motivation behind this initiative is that the primary obstacle in the development of AVs is no longer technological; rather, it is the availability of robust data. According to Naga, companies like Waymo face challenges in accumulating the necessary data to train their models for specific scenarios in various locations. Many AV companies lack the capital to deploy fleets to gather essential information, highlighting a significant market opportunity for Uber.
Positioning itself as a data provider within the AV ecosystem is a strategic move for Uber, particularly after it previously stepped back from its self-driving car ambitions—a decision that some, including co-founder Travis Kalanick, consider a critical error. Currently, Uber collaborates with 25 AV companies, including London-based Wayve, to construct what Naga described as an “AV cloud,” a repository of labeled sensor data accessible to partner companies for model training. Moreover, these partners will be able to run their AI models in a ‘shadow mode’ using actual Uber trip data, simulating AV performances without the need to deploy vehicles.
Naga emphasised that Uber’s intention is not to monetise the data but to democratise access to it. However, given the significant market value of the resources Uber is developing, it remains to be seen whether this altruistic positioning will endure. The company has made investments in various AV firms, and the capability to supply proprietary training data on a grand scale may bestow considerable leverage within a sector that presently relies on Uber’s ride-hailing platform to connect with customers.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
