Xpeng shares progress of autonomous driving base model research and development at AI conference CVPR

Beijing time , the top computer vision conference CVPR 2025 was held in Nashville, Tennessee, USA. Xpeng Motors was invited to participate in the CVPR WAD (Workshop on Autonomous Driving) seminar, and discussed the latest AI technologies in the industry with autonomous driving peers from industry and academia such as Waymo, NVIDIA, University of California, Los Angeles (UCLA), and University of Tuebingen.

Since the AI ​​big model wave, the field of autonomous driving has undergone a technical paradigm shift, from the past model of human handwritten rules to the AI ​​model trained based on massive data. The relevant technical progress has also become a hot topic at CVPR in recent years. At this year’s CVPR WAD, Dr. Liu Xianming, head of XPeng’s World Foundation Model, delivered a speech entitled “Scaling up Autonomous Driving via Large Foudation Models”, systematically introducing the process and methods of XPeng’s self-developed industry’s first ultra-large-scale autonomous driving foundation model, and disclosing its cutting-edge exploration in model pre-training, reinforcement learning, model vehicle deployment, AI and data infrastructure construction, bringing practical experience in the industrial field to peers.