At CES 2025, Uber teams up with Nvidia to scale autonomous driving faster

At CES 2025, Uber teams up with Nvidia to scale autonomous driving faster | TechCrunch

Uber has partnered with Nvidia to leverage Cosmos, a generative world model, and DGX Cloud for developing autonomous vehicle technology. By utilizing physics-based simulations trained on vast video datasets, the collaboration aims to create realistic driving environments for training AI models. This strategic move allows Uber to accelerate the deployment of self-driving solutions without directly manufacturing the underlying hardware or software, relying instead on high-performance cloud infrastructure to refine its algorithms efficiently. The partnership addresses Uber’s historical struggles with autonomous development, following legal disputes and safety incidents that led to the sale of its self-driving unit. Rather than building technology in-house, Uber focuses on being a logistics bridge between riders and drivers, including robots. By integrating Nvidia’s tools, the company hopes to overcome previous setbacks and scale its operations more rapidly, capitalizing on its extensive network of partnerships with other autonomous firms to integrate self-driving vehicles into its service offerings. This initiative is highly relevant to open_data because it underscores the critical role of massive, diverse datasets in training robust AI systems. The reliance on billions of video tokens highlights the industry's need for comprehensive, high-quality data sources to simulate real-world scenarios safely. As companies like Uber and Nvidia push for faster autonomous scaling, the implications for data privacy, copyright, and the accessibility of training data become increasingly significant for the open_data community.

Source: techcrunch.com
Published on 2025-01-08

Related news

« Australia's housing shortage deepens

Arranca certificación de motociclistas en Edomex | El Universal »