美股IPO·2025-10-23 23:28Core Viewpoint - Nvidia is collaborating with Uber to utilize diverse driving scenario data collected by Uber to enhance the training of Nvidia's Cosmos World foundational model for autonomous driving technology [1][3][4]. Group 1: Collaboration Details - Nvidia Drive announced the partnership with Uber, focusing on leveraging Uber's extensive real-world driving data to accelerate advancements in autonomous driving [3][4]. - The collaboration aims to achieve three main technical objectives: higher precision in simulation, reduced iteration cycles for model training, and improved stability in rare or extreme scenarios [5][9]. Group 2: Data Advantage - Uber's global operational network provides unique data value for training autonomous driving models, with over 1 billion rides and deliveries conducted monthly [8]. - Uber vehicles operate more frequently in complex situations, such as adverse weather and crowded events, which are less common for private vehicles, thus enhancing the data collection for rare scenarios [8][10]. Group 3: Specific Use Cases - For instance, Uber's airport pickup and drop-off services operate under various weather and lighting conditions, making it challenging to sample these scenarios adequately in other contexts [13]. - The unique environment of airport pickups, characterized by dense traffic and unpredictable pedestrian movements, presents specific challenges that can be effectively captured through Uber's data, including driver performance metrics [14].