When sales competition can no longer meet the increasingly fierce market competition in the automotive industry, He Xiaopeng, Chairman and CEO of Xiaopeng Motors, began to call for higher quality internal competition in his speech.
"Americans are thinking about how to make the end-to-end automatic driving assistance technology better and how to do it better than people. While Chinese technology companies are still 'rolling' weekly sales charts, trying to make money." He Xiaopeng said frankly, "This is not the way science and technology competition should be."
'End to end', a technical term that somewhat confuses consumers, has become another battleground for the 'intelligent' competition in the automotive industry.
Northeast Securities Research News shows that the end-to-end large model is the third stage of the development of the auto drive system. It can directly realize the processing from perception to control by integrating the perception, decision-making, control and other module functions into a unified framework.
In other words, in the past, intelligent driving systems for automobiles were divided into multiple modules, where vehicles "saw" environmental information through cameras and radar, understood scene content, predicted future developments, decided on driving trajectories, and ultimately controlled the vehicle's movement through different steps on a "production line". And 'end-to-end' can directly complete the overall process from the raw data input end to the final control output end, eliminating the loss in information transmission and greatly improving the efficiency of intelligent vehicle data processing, thereby improving system response speed. It is regarded as the mainstream technology route for achieving fully autonomous driving in the industry.
In March 2024, Tesla began to launch FSD v12 on a large scale in North America. This end-to-end intelligent driving system has received widespread attention in the market and stimulated domestic car companies to accelerate the research and promotion of end-to-end technology.
Recently, several emerging car companies have announced progress in the field of end-to-end large-scale models for autonomous driving.
On July 5th, Ideal announced the launch of a nationwide map free NOA and the release of a new autonomous driving technology architecture based on end-to-end models, VLM visual language models, and world models to all Ideal AD Max users.
On July 27th, NIO officially released its intelligent driving technology architecture and stated that it has upgraded the algorithm layer by introducing its self-developed NIO World model end-to-end architecture, making the intelligent driving experience safer and more human like.
On July 30th, Xiaopeng Motors announced that it will fully promote the AI Dimensity System XOS 5.2.0 version to users worldwide. He Xiaopeng emphasized that Xiaopeng Motors maintains a leading position in the domestic market in end-to-end large-scale production of vehicles. With the help of end-to-end big models, the evolution cycle of Xiaopeng's intelligent driving technology and experience will be significantly shortened.
On July 4th, Gu Junli, Deputy General Manager of Chery Automobile Co., Ltd. and General Manager of Dazhuo Intelligent Technology Co., Ltd., also stated that currently, autonomous driving has fully entered the industrialization stage, and end-to-end is the "inevitable battle" under the new generation of technological revolution. As a brand of Chery Automobile's autonomous driving technology, Dazhuo Intelligent Technology is expected to achieve end-to-end large-scale production on high-end models by 2026.
Guosen Securities stated that the current automotive intelligence has achieved a technological breakthrough in end-to-end large-scale models, and the domestic market has a market space of nearly 300 billion yuan in the long term. The singularity of intelligent driving is approaching.