Automobiles are moving from the industrial age to the digital age, and autonomous driving has become the undisputed future of the industry. Since 2019, when Tesla launched the competition to upgrade the intelligent driving experience with high-speed leadership, it has gone through four years from Weilai, Xiaopeng to Geely, BAIC, and Great Wall. With the rapid development of new energy technologies, intelligent competition It is becoming a "battlefield that determines the future life and death" of major new and old automakers at home and abroad.

At present, in the field of passenger vehicles, the first half of the autonomous driving technology is still anxious. Almost most of the models sold by mainstream automakers are transitioning from L2 to L3+. When the autonomous driving competition comes to the second half, what kind of key capabilities can drive autonomous driving to a high degree or even complete automation?

Grasp the "decision-making core" in the technical architecture

As we all know, the development of autonomous driving is inseparable from the three core technical frameworks of "perception, decision-making, and execution". The execution layer is the easiest for users to understand. It drives the vehicle to execute corresponding command actions through the chassis and various control accessories. It is closely related to the ultimate application experience, which is not difficult for most car manufacturers.

As we all know, the development of autonomous driving is inseparable from the three core technical frameworks of "perception, decision-making, and execution". The execution layer is the easiest for users to understand. It drives the vehicle to execute corresponding command actions through the chassis and various control accessories. It is closely related to the ultimate application experience, which is not difficult for most car manufacturers.

In fact, in these three closely connected links, perception ability is the foundation, decision-making ability is the key, and decision-making ability is deeply dependent on computing power and storage power. The decision-making link is like the "brain" of the car. It not only needs to receive and store the environmental information and vehicle positioning provided by the perception layer, but also needs to integrate analysis, training and simulation based on a large amount of data accumulation through algorithm fusion and feature extraction. And give the best matching decision, output to the control units of various execution layers.

It can be seen that behind the increasingly fierce competition for smart car autonomous driving, the root cause is actually the competition between computing power and storage power. What is certain is that car companies with long-term plans in the field of autonomous driving, whether they are "new forces" or "old brands", are constantly making long-term deployments with higher storage and computing power, from the cloud to the intelligent computing center, in order to cope with Sustained surge in demand for computing power challenges, and long-term cost control to maintain the core competitiveness of its own development. In August 2022, Xiaopeng Motors built the "Fuyao" intelligent computing center, with a computing power of 600PFLOPS. In January 2023, Momo Zhixing, an autonomous driving company under Great Wall, established the Snow Lake Oasis Intelligent Computing Center, with a computing power of 670PFLOPS. In the same year, Geely The car builds the Star Wisdom Computing Center, with a computing power of 810PFLOPS... It can be said that higher-level computing power has become "a battleground for military strategists"!