High-end intelligent driving, officially launched the standard offensive.
Analogous to the smartphone process, flagship models have entered the stage of popularization for everyone:
Whether it is autonomous parking, high-speed loop NOA, or memory navigation for urban commuting, it can recognize traffic lights, avoid pedestrians, autonomously bypass obstacles, and turn left and right, and the active safety functions with a full score of five stars in CNCAP,...... It should be fully matched.
And this plan is also aimed at fuel vehicles, shouting fuel vehicles, and the topic of the Chinese car circle where oil and electricity are opposed.
The most important thing is that such a high-end intelligent driving product, the experience is going up, and the price and cost are constantly falling. Thousands of yuan hardware cost, the standard configuration is the flagship high-end intelligent driving ability......
搅动风云者:高通、Momenta。
Two of the best companies in their respective fields have rolled up in a new direction.
Qualcomm Momenta has teamed up to make the world premiere of the latest generation of the Snapdragon Ride platform
The two sides have just officially announced that the new intelligent driving solution is based on Qualcomm's latest Snapdragon Ridep platform (Snapdragon SA8620P and Snapdragon SA8650P):
Among them, the Snapdragon SA8620P chip is manufactured on a 4nm process and has 36TOPsAI dense computing power, which is more friendly to Transformer architecture support. Transformer is the technical basis for realizing BEV perception capabilities, which is high-end cities and high-speed NOAs.
In terms of sensor solutions, you can choose to carry two different solutions: 7V3R and 7V1R. It is suitable for mainstream passenger cars of 10-200,000 yuan.
Note that it is a "mainstream passenger car", and intelligence no longer distinguishes between oil and electricity.
According to reports, this is because the Snapdragon 8620 chip can support passive heat dissipation, and for fuel vehicles with highly mature infrastructure and low "redundancy" of on-board electricity, they can be directly put on the car without adding additional costs.
In terms of specific functional experience, it realizes "integrated driving and parking", including L2 ADAS full functions, and can also support high-end intelligent driving functions such as HNP high-speed elevated pilot assist (high-speed NOA), MNP memory navigation assistance (commuting NOA), LPNP memory parking pilot assist, PNP parking pilot assist, etc.
Momenta CEO曹旭东认为:
"The Snapdragon 8620 chip provides a new possibility to achieve more cost-effective high-end intelligent driving solutions, and Momenta's algorithm advantages can be further leveraged to accelerate the implementation of high-end intelligent driving in more mainstream price segments.
8620 is the hardware foundation of "oil and electricity with intelligence".
In addition, Qualcomm Vice President Xian Lei said:
"Momenta's algorithm capabilities and Qualcomm's technical advantages in chips are a combination of two swords, which can bring a better product experience, more friendly costs, and a wider penetration rate of intelligent driving.
Momenta's algorithm capabilities and data closed-loop "flywheel" are standard software guarantees for high-end intelligent driving.
Both sides mentioned the popularity and standard configuration of high-end intelligent driving. This is actually one of the cores of this year's intelligent driving competition.
There are two important indicators, one is to take the high-speed NOA that can cover the whole process from the toll station to the toll station, and can change lanes, avoid, and enter and exit the ramp autonomously as the starting point of the functional experience.
The ACC that did not have a navigation function before, or the assisted driving function that could only realize the "lever to change lanes" is no longer competitive.
The second indicator is that the high-speed NOA function no longer costs tens of thousands of yuan to choose, and the standard configuration with the car and the model has crossed the watershed of 150,000 yuan.
Qualcomm and Momenta's new solution is the industry's first product with a dedicated strategy and clear configuration for high-end NOA as standard.
The speed is amazing.
According to reports, the two parties have already discussed the feasibility of cooperation as early as the chip research and development stage, and give full play to Momenta's experience and advantages in mass production and delivery. It only took less than two months to run through the product functions.
It should also be the world's first Refcar based on a production car.
In fact, this is also Qualcomm, the king of the smart cabin, officially opened the popularization war of intelligent driving, and it was further revealed that the cooperation is not only Qualcomm's Snapdragon 8620 chip, but also includes Snapdragon 8650 - which means that the most advanced high-end intelligent driving capabilities such as urban NOA and memory routes have also started mass production and getting on the car.
This is the foundation of Qualcomm's "hard" strength, and it is also the speed of the soldiers under the flywheel drive of Momenta.
Impact: Joining forces with high-end intelligent driving to enter the "standard era"
The value of the first and second car is not only for Qualcomm and Momenta, but also for the automotive industry and all potential consumers.
There are at least three clear points.
First, break through the barrier of "less than 150,000 without intelligence".
Second, the intelligent experience no longer discriminates between oil and electricity.
Third, for consumers, the popularization of intelligent cars has entered the era of "thousand-yuan flagship machines" analogous to smartphones.
First, for the first time in history, an entry-level model enjoys the same intelligent driving experience as a luxury model.
In the early stage of the development of smart cars, the cost of high-speed NOA hardware configuration is very high. Just equipped with a domain controller that can support high-end intelligent driving chips, the cost will cost thousands of yuan or even more than 10,000 yuan, plus a full set of sensors for lidar, camera, millimeter wave radar... It's easy to exceed 10,000 yuan, and the hardware cost alone has blocked many mainstream models of 10-200,000 yuan.
At that time, entry-level models below 150,000 yuan were equipped with high-end intelligent driving, and the price would rise to 200,000 yuan, losing competitiveness.
However, after several years of penetration of smart cars into traditional cars, the scale of the supply chain has expanded rapidly, and the cost reduction effect is obvious.
Qualcomm's Snapdragon 8620, Snapdragon 8650... A series of vehicle-grade intelligent driving chips can be said to be at the right time in the cycle from the launch to the landing of mass production models, taking advantage of the latecomer.
This allows Qualcomm and Momenta to reduce the BOM cost of the new platform to the order of several thousand yuan - the price of a flagship smartphone is reasonable enough for a model priced under 150,000 yuan, and there is no reason not to come standard with the car.
The second point is that the majority of fuel vehicle users are finally no longer forgotten in the wave of intelligence.
From a technical point of view, the electronic and electrical architecture of oil vehicles is mature, but it also represents a lack of support for intelligence, such as the delay of signal transmission, the problem of redundancy, and the problem of slow control feedback response...
Therefore, the core problem of intelligent oil vehicles is the integration of intelligent driving hardware in the architecture and the improvement of response speed in algorithms.
At the hardware level, Qualcomm brings the advantages of high performance, low power consumption and passive cooling. Momenta replaces rule-driven with data-driven, and in the face of the characteristics of different power systems, it trains AI to control the accelerator and brake with the method of mature human drivers to the greatest extent, avoiding complex manual adjustment and adaptation.
Taken together, the standard configuration of high-end intelligent driving benefits the two largest user groups.
One is the users of fuel vehicles who account for half of the new car sales, and the other is the users of 10-200,000 models who also account for half of the share.
There is no range anxiety, high-end intelligent driving is standard with the car, the price is flat and the experience is far better than that of traditional "function cars".
This kind of meaning and value is difficult not to be compared to the wave of popularity of smartphones, the price of thousands of yuan but with some of the functions and features of flagship smartphones, such a driving change - truly let everyone enter the wave of smart life of the mobile Internet without discrimination.
For mainstream passenger cars, the significance is not only the comfort of the experience, but also the ability to choose the once exclusive luxury car, which is now accelerating the standard configuration to every vehicle and empowering every car owner.
More importantly, the popularization of intelligent driving and intelligent vehicles will definitely promote the entire road conditions and traffic to a new stage of safer.
This is a boon for all users and potential consumers.
High-end intelligent driving has entered the "standard era", what else does it reflect?
In fact, the historical process of the "standard era" of high-end intelligent driving has come rapidly, but not without warning.
At the scene of the cooperation release, Xian Lei, vice president of Qualcomm, and Cao Xudong, CEO of Momenta, shared the law behind the action in the conversation.
The first is Moore's Law of AI.
The second is the flywheel effect that Momenta has been practicing.
The so-called AI Moore's Law, Momenta CEO Cao Xudong recalls the industry trend 2 years ago to let more people perceive:
Just two years ago, intelligent driving was still competing for the mass production of high-speed NOA, and most of the product experience was not up to par, and the scene of making people break out in a cold sweat often happened.
Now the high-speed NOA has completely covered all scenarios from toll stations to toll stations, and has dropped from 300,000 models to 200,000 models.
In 2 years, high-speed NOA has changed from a "touchstone" for intelligent driving to a "basic compulsory course".
Cao Xudong judged that the experience brought by AI technology and products will be improved by 10 times every two years - this is an insight from the current front-line practice, but it does reflect the development trend of the industry.
Therefore, the CEO of Momenta believes that in another 2 to 4 years, the penetration rate of high-end intelligent driving will exceed 50% and begin to become the majority.
Qualcomm has completely experienced the development of PC, Internet, mobile phones and other smart terminal categories in various eras, and from the perspective of its own experience, Qualcomm also agrees with Momenta's insight, believing that every era has a cycle of rapid technological development and rapid iteration of commercialization.
And artificial intelligence is now in such a detonation period.
Behind Moore's Law of AI, it is actually data that plays a further role.
Cao Xudong emphasized that the autonomous driving industry has obvious data-driven characteristics. The processing and training of massive amounts of data can greatly improve the performance of the system, which means that the need for sensors can also be reduced.
That is to say, from the perspective of cost, AI Moore's law can be expressed as: "The cost of intelligent driving products with the same performance needs to drop significantly every two years".
It is precisely this kind of insight from data iteration that has been used by Momenta as a guide, methodology, and winning weapon before Momenta.
The so-called flywheel effect simply means taking data as the core, building closed-loop automation of data iteration, and then driving the end-to-end improvement between L4 technology flow and L2 mass production data flow.
Once L4 high-end technology is empowered to L2 mass-produced products, and the massive data generated by mass-produced vehicles is returned and the algorithm is automatically trained, the capabilities of AI drivers will improve rapidly.
With the continuous expansion of mass production and installed capacity, the "flywheel" will be able to spin faster and faster, the intelligent driving experience will be better and better, and the technical barriers of Momenta will be higher and higher.
So this can also explain why the "standard era" of high-end intelligent driving will come here and now, and Qualcomm and Momenta will take the lead in firing the first shot.
This is inseparable from Qualcomm's accumulation in chips, computing and underlying AI software and hardware architecture, and the balance between cost and performance of smart cars, including solving the heat dissipation problems that need to be overcome for fuel vehicles, and laying the foundation for explosive models.
It is inseparable from Momenta's skill, efficiency and reputation in high-end intelligent driving, especially the technology and product iteration results brought about by the flywheel effect, so that inside and outside the industry, up and down the supply chain, can truly believe because of what they see.
Interestingly, before Qualcomm and Momenta joined forces, the outside world was more about the standard configuration and popularization of high-end intelligent driving, and it was more hoped that after the fuel vehicle was completely "new energy", it would be electrified first, and then intelligent.
However, the two best companies in their respective fields have changed this process, whether it is AI or high-end intelligent driving, they should not be divided according to the form of energy and power, let alone in price.
AI empowers everyone, AI empowers every vehicle, and drives a safer, more convenient, and more efficient future...... This is the most simple and valuable meaning of the "standard era" of high-end intelligent driving.