NIO Officially Forms Large Model Team to Accelerate the Landing of Urban NOA
One of the new forces in Chinese production, NIO, has recently completed a restructuring of its research and development department. The original “perception” and “regulation control” teams have been merged and will be collectively referred to as the “big model team” for future management; the original “integration” team has been reorganized into a delivery team with no change in operating mode.
In terms of management, after the restructuring, the research and development department is still led by Vice President of Research and Development Ren Shaoqing. The merged big model team is now led by Peng Chao, former head of the perception team.
According to public information, Ren Shaoqing graduated from the joint doctoral program of the University of Science and Technology of China and Microsoft Research Asia. He previously served as the Director of R&D and Co-founder at Momenta before joining NIO in August 2020, already being a prominent figure in the international computer vision field.
Peng Chao graduated with a master’s degree from Tsinghua University and previously worked as a Senior Visual Algorithm Engineer at Momenta.
After this business adjustment, Ren Shaoqing stated that he would abandon the traditional paradigm of “perception-decision-control” that has been used in the industry for many years to develop intelligent driving. In the future, they will gradually transition to an end-to-end large model paradigm and use deep neural network technology to achieve advanced intelligent driving.
Analysis indicates that in the traditional intelligent driving research and development model, perception, decision-making, and control are usually divided into different independent modules. When facing complex road problems, there may be situations that cannot be handled. The end-to-end large model can avoid such problems: unlike traditional modular autonomous driving systems, the end-to-end large model is a whole entity that can cover a wider range of driving scenarios through deep learning technology and accelerate the implementation of advanced intelligent driving functions such as urban NOA.
It can be said that large-scale technology is the future of intelligent driving for car companies. Currently, XPeng Motors and Li Auto have independently incorporated large-scale technology into their own intelligent driving systems: Li Auto not only independently developed the multimodal cognitive large model Mind GPT but also jointly established an Intelligent Computing Center with Volcano Engine; XPeng has implemented end-to-end intelligent driving large models and launched the AI Tianji system. In addition, Nezha Motors, GAC Group and other car companies are also actively deploying in the field of on-board large models.
SEE ALSO: The World’s First NIO Fourth-Generation Power Swap Station Goes Online
On April 12th, NIO announced the official launch of NOMI GPT push. NIO stated that after users successfully upgraded the NOMI GPT function, they can experience a variety of new interactive experiences including large-scale encyclopedia, unlimited chat, magical atmosphere, fun expressions, car Q&A, AI scene generation and more under the empowerment of intelligent AI for a brand-new travel experience. Since its release, NOMI has undergone over 130 version iterations and added or updated more than 2,000 functions.
On April 30th, NIO pushed the full-domain Navigate on Pilot (NOP) + urban function to all Banyan·Rong intelligent system users. This is the largest-scale urban intelligent driving push in China so far. According to official data, as of the day of the press conference, NIO has conducted sufficient road verification for NOP+ across domains, with a total road verification mileage of 1,207,977 kilometers covering 726 cities. Among them, highway and urban expressway verification mileage reached 360,000 kilometers and urban road verification mileage reached 847,000 kilometers.
William Li, Chairman and CEO of NIO stated that NIO‘s Intelligent Driving goal for 2025 is for autonomous driving usage time to reach 80%, which will be ten times safer than human driving.
