Our Science innovation is dedicated to developing foundation models for autonomous driving. These ‘general purpose’ models excel at various driving tasks such as perception, reasoning, and motion planning.

We train these models using an extensive corpus of diverse vision, language, and action data from our R&D fleet, partner fleets, and our generative AI models. This multifaceted training approach provides a deep and nuanced understanding of the dynamics of the physical world and fine-tunes the model’s decision-making for safe driving on the road.