Text: Tan Qing said AI Author: Zheng Driving
The annual Consumer Electronics Show (CES) arrived on schedule.
The 2019 of autonomous driving is actually not good, that is, the track is clamped by factors such as high costs, the technical bottleneck began to highlight the year, and the capital winter followed, but the turning point occurred at CES in 2020.
In 2020, CES in the field of automatic driving, is the perception equipment manufacturers C position of the first session, from the domestic Huawei, DJI, to foreign mechanical lidar originator Velodyne and other companies have come up with cheap lidar equipment, compared to 75,000 in 15 years and often out of stock mechanical lidar, the industry can be described as ushering in a dawn.
The first step to doing autonomous driving is undoubtedly to survive, and after realizing this, many players have let go of the obsession of climbing the peak towards L5 technology.
In 2020, the tears of the perception device sale made everyone throttling, and 2021 is the open source year of automatic driving, with the scene as the entry point to land technology, in this year has gradually become the consensus of the industry.
At CES 2021, NXP highlighted the new ADAS solution, starting with advanced assisted driving in passenger cars that are easier to land; Mobileye also followed the lidar trend that more OEMs began to favor, launching a lidar system integration chip, while expanding the test fleet on a larger scale.
Everything seems to be laying a solid foundation for landing.
Over the years, CES is often a time when major head players on the technology subdivision track compete to show their muscles, which can also reflect the evolution of many subdivided tracks in the next year or more.
This year's CES, which also heralds autonomous driving in the next few years, seems to have begun to face a new proposition for the next stage.
From CES, look at the next hill of autonomous driving
At this year's CES exhibition, the automotive industry players have thrown out a lot of heavyweights, one of the most explosive areas, is undoubtedly the main engine manufacturers competing for glory, not only BMW released a quite eye-catching "electronic ink" technology, can support the body to achieve "one-click color change" function, Mercedes-Benz, Cadillac, Volvo and so on have also released new concept cars.
Even Sony couldn't hold back, and quite ironally joined the already extremely crowded car-making army at this juncture today.
However, OEMs spend a lot of time, and it is difficult not to pay attention to the star segment of the auto industry, which is autonomous driving. The first is that mobileye in the self-driving team finally stopped "obscene development" and showed off some muscles at the meeting.
As a veteran self-driving technology and chip research and development company, Mobileye has been favored by Intel in 2017 due to the rapid growth of chip shipments and excellent market performance, but after being acquired by Intel, it quickly learned the traditional art of "the owner" - "squeezing toothpaste".
When Nvidia's latest autopilot chip Atlan hashrate reached a terrifying 1000TOPS in 2021, and Tesla's FSD chip hashrate reached 72TOPS, as the latest EyeQ5 chip of Mobileye, which was once a leading technology, the computing power is only 24TOPS, and the computing power is 41 pieces of the top people Atlan...
But then again, Mobileye's strength is not in the computing power but in the algorithm, taking the weaknesses of others and the strengths of opponents, it is not fair, and this year's Mobileye seems to have suddenly stood up, not only released the EyeQ Ultra chip that directly matches Nvidia Orin, but also set a small goal to put into production in 2024 L4 level autonomous vehicles for the C-end consumer market that cooperate with Extreme Kr.
Coincidentally, GM also set a somewhat similar flag at this year's CES to offer self-driving (at least L4 level) cars for individual users by the middle of this decade.
From this year onwards, in two or three years from the production of L4 self-driving cars sold to individual consumers, if a company makes this bold statement, it will be doomed to be a little eye-catching suspicion, but the two companies invariably set up a similar flag, it is worth making us think deeply.
We must first understand how meaningful this flag is.
L4 until today, has been the main battlefield of Waymo, Baidu and other technology companies, although Tesla has more than once claimed that its hardware technology can support L4 technology implementation, but from a practical point of view, there is not yet a L4 car in the world that can be sold to C-end consumers.
Even the L3 car, with the exception of mercedes-Benz, which was internationally certified to be on the road only in Germany at the end of last year and has a speed limit of 60 km/h, can hardly be found in other cases in the world. Therefore, at this year's CES, the forward-looking and heavyweight of these two flags is self-evident.
Although the flag was erected, Xu Hang (pseudonym), as an industry insider, expressed his own views to us on this matter:
"Cars with open L4 technology want to be sold to the C-end, and it is two basic elements that can be achieved to solve long-tail safety problems such as The Corner Case as soon as possible in the next few years, and further pull the cost of the vehicle to a controllable range." After all, only by passing the two major barriers of policies and regulations and controllable prices can L4 cars obtain the most basic scale commercialization possibilities. ”
Starting from this basic principle, we may wish to speculate on what the new wind direction of the autonomous driving track will be in the next few years.
At 22 years of CES, compared with the past, a somewhat different point is that the faces of OEMs such as Extreme Kr and GM have begun to appear more and more in the automatic driving track, and therefore, Tan Qing said that AI believes that in the next few years, cooperation may become a new normal in the autonomous driving track, and around this to promote the further and diversified landing of automatic driving technology, it will become a new trend in the track.
From the arms race to the maximization of efficiency
We all know that on the autonomous driving track, data is often one of the most important core assets of various parts of the enterprise.
"When autonomous driving first got up a few years ago, everyone was very sensitive to data and was unwilling to give in on data. Technology companies are reluctant to give data to OEMs, and OEMs are reluctant to give data to technology companies. "An insider of an autonomous driving company once told us.
"But you will find today that if the companies in all parts of the track are still deadlocked, it is difficult for the entire industry to take a step forward." So for things like testing and operational data, the current (industry) is already moving towards a more open and shared way, as long as it meets the requirements of policies and regulations, and then it is beneficial to both sides, and many companies are willing to do it. ”
As he said, compared with the more closed-door industry environment on the track in previous years, the cooperation between all parties has become a trend to promote intelligent development, in fact, from the perspective of home and abroad, this trend is also common.
Take Geely, for example, at the end of last year, Waymo, one of the oldest companies on the autonomous driving track, announced that it will carry out in-depth cooperation with Geely in the field of Robotaxi, and Geely will create the first Robotaxi model designed for Waymo. A few years ago, the case of OEMs building custom cars for self-driving companies was almost non-existent.
In addition to Geely's cooperation with Waymo and Mobileye, Xiaoma Zhixing has cooperated with FAW (Nanjing), Bosch and other enterprises, and the iron triangle model built by Wenyuan Zhixing and GAC and Ruqi Travel has been announced or launched last year.
It is not difficult to find out that compared with previous years, the autonomous driving track has suddenly turned to a new industry model of "cooperation is king", and what kind of changes has it undergone in the underlying logic?
"Cooperation is actually to accelerate the landing, but why is the landing of automatic driving still a problem after so many years?" The main reason is that there is a particularity in the industry. Industry analyst Ye Jun (pseudonym) told Tan Qing that AI said, "Autonomous driving is a technology-intensive track, and the technical reserves in the early stage or your entrepreneurial technical team lineup are very important, in the final analysis, it is because everyone wants to build their own technical barriers in the early days." ”
"In contrast, in the hundred groups or the online car war of that year, this difference is reflected in the obvious, and the speed of people in the education market in that year was really beyond the reach of automatic driving." This is mainly because (online car, local life) the track is not a scientific and technological barrier, up is to grab the market, the so-called fight is good, incidentally can efficiently educate the market, forcing the regulations to land. ”
"Now the automatic driving for the large-scale C-end of the road policy only open L2, so that high-end technology landing, must be inseparable from the acceleration of the education market and forced the landing of regulations, now the technical bottleneck highlights the financing is not easy to engage, and then closed to engage in inevitable repeated wheels, the industry to reach a consensus on cooperation can actually be said to be an inevitable product of this stage."
Tan Qing said that AI believes that as Ye Jun said, automatic driving is entering a new stage, but this new stage does not mean that track players should start to engage in technology monopoly and grab the market, the main task orientation of the next stage is actually to use cooperation to accelerate the landing to overcome the long tail problem, with safety and scale to activate the market first, and then the market competition stage, openness is an essential element of the current track.
Maximizing efficiency doesn't stop at "small circles"
The car that opens up L4 capabilities to the C-end is sold for the first time this year by Extreme Kr and GM, and in the case of Extreme Kr, we can better understand the overall track direction in the future by studying the strategic logic of the flag proposer.
In fact, in the field of automatic driving, Geely's layout is not weak. Previously, Geely and Baidu jointly established a new brand of Jidu Automobile, tianyancha APP shows that Jidu Automobile Co., Ltd. was established on March 2, 2021, so far, Geely has Baidu in the field of automatic driving, an ally with a leading position in the industry.
In addition, Geely also has Li Shufu as early as 2016 founded the automotive intelligent technology company YiJiatong, Yijiatong and invested in the establishment of core Qing Technology, in general, the main business position of the two companies is in the car chip, the main business model is that The core Qing Technology is responsible for solving back-end problems such as power consumption, working conditions and computing power, and Yi Jia Tong Technology is responsible for solving front-end problems such as product perception, user experience and system.
It is not difficult to find that in the smart cockpit and automatic driving chip, Geely can be said to have a hands-on attitude to a certain extent, at the same time, Baidu is also a close friend of Geely, not only the two companies jointly established the Jidu brand, Yijiatong also has Baidu's A round of 1.3 billion yuan to lead the investment.
In the automotive intelligent warfare, Geely has experience in automobile manufacturing, Baidu is an ally in automatic driving, and there is a billion coffee pass in the development of automotive intelligent chips and systems, and the existing iron triangle model, Geely also cooperates with mobileye this time, which is inevitably surprising.
But in fact, from this new cooperation, we seem to be able to infer the three logics revealed by Geely for the "abandonment" of its own iron triangle by the extreme Krypton brand.
First of all, Baidu, after Geely announced the cooperation between the Extreme Kr brand and Mobileye, some voices on the Internet pointed out that the cooperation between Baidu and Geely may have problems, otherwise why would Geely find another partner for the Extreme Krypton brand?
In fact, this somewhat "conspiracy theory" implication of this argument cannot withstand careful scrutiny. We all know that Baidu has always been in the first echelon of autonomous driving technology companies in China, and the reputation of "Autonomous Driving Whampoa Military Academy" and "China Waymo" is not empty, but the technology it is good at is more at the software level.
For example, at this year's CES, not only Did Extreme Krypton announce cooperation with Mobileye, but even Jidu itself announced that the new car will cooperate with Nvidia, equipped with Nvidia's Orin chip.
Tan Qing said that AI believes that the cooperation between Extreme Kr and Mobileye should not have a substantial impact on the cooperation between Baidu and Geely, but for Baidu, compared with Huawei, which also provides automatic driving technology, its current foundation is relatively weak in terms of sensor hardware such as autonomous driving chips and lidar.
As Huawei continues to grow its circle of friends, Baidu also needs to think more about the new trend of embracing cooperation in the future of autonomous driving.
For YiJiatong, this extreme krypton cooperation with Mobileye, which is mainly to make chips, even the "pro-father" does not choose Yijiatong, who is also the "pro-son" of chips, which seems to let people see that Yijiatong has not yet reached the technical reserves recognized by Geely in the chip research and development and system architecture of automatic driving.
Along with this judgment, a careful search can be found that although the current Yijiatong is said to exist in the chip business, the product that really makes it out of the circle is actually a 7nm-level Dragon Eagle No. 1 intelligent cockpit chip and the intelligent cockpit system Geely Galaxy OS, which does not seem to have too many products in the field of automatic driving chips.
When this point is fed back to the strategic decision made by Geely for the Extreme Krypton brand, in tan Qing's AI view, Geely should not care too much about the "body and soul theory" preached by SAIC, but chooses to use the efficiency maximization method to layout the overall business, which is not limited to all technologies to engage in full-stack self-research, and is not limited to just building a "small circle" closed loop.
Overall, at this year's CES, Extreme Kr dared to release the L4 level autonomous vehicle that will be put into production in 2024 for the C-end consumer market, which makes us see the strategic logic behind it, and has its own unique insight into the word "cooperation" in the field of automatic driving.
The insider of the self-driving enterprise mentioned above told us, "Our current cooperation is also advocating openness, not exclusivity, which is true for every of its partners, because at this stage of development, only by involving more diversified enterprises to jointly do a big industry is the most critical." ”
The new hill that the autonomous driving industry will face in the new year may be a group game, but how many teams can cross it? We'll see.
Ps: Tan Qing said AI, turn left new energy vehicles, turn right smart driving, there is depth, there is temperature, the author Zheng driving, reprint please retain copyright information.