During the auto show, there are many big news on automatic driving, there are many topics worthy of further discussion, but in order to avoid high-intensity information bombings in the past week, “Nine Chapters” has chosen “wrong peaks”.
The topics involved in this article are:
1. “Traditional IT Think” VS “Internet Think”;
2. After Huawei “Bright Sword”, Robotaxi or will be forced to adjust the strategy;
3. “Huawei Inside” becomes “Huawei Outside”, the Biece of To B and TO C is blurred;
4. Huawei has no external imagination, “strong”;
5. Advanced automatic driving begins to work toward “liberation of eyes”;
6. There must be “soft and hard integration”, but also have a “hardware and software decoupling” attitude;
7. The two trends in the stack and hardware reduction are parallel;
8. Is the laser radar really just “redundancy”?
9. The decision algorithm is not good, and the plurality of sensors are useless;
10. The low line number laser radar market will be eaten;
11. Daily business must accelerate the collection of high-precision maps of urban roads;
12. “Little Business” in the big scene and the “big business” in the small scene;
13. Do not simply put the smart car ratio as a smartphone;
14. Those companies that “lack aura” may need to be revalued;
15. “High-end” may become an disadvantage of Apple.
Strictly said that this is not a “article”, which is a “article”, but is some “Donghua West”, but we believe that many phenomena and perspectives mentioned in the article. There is still a big discussion value.
One. “Traditional IT Think” VS “Internet Think”
Huawei, the person in charge of the ADS, said that many words that were reported when they were accepted by the media group, which led to a lot of “Huawei so radical” in the outside world; but there have been some automatic driving companies. In the eyes of contact, Huawei is “not radical”.
After the MDC 810 was released, a friend who focused ADAS’s Tier 1 said on WeChat: “I used to be an IT industry, and I jumped to automatically driving this year. There is an obvious feeling. The automatic driving industry is generally biased. The way of thinking, even if there is no need to promote, so most of the parameters are on paper; compared to the traditional IT industry thinking, the development is 60% or more. “
two. After Huawei “Bright Sword”, Robotaxi will be forced to adjust the strategy
Objectively speaking, from the perspective of the user experience, Huawei and North Automotive Fox show automatic driving ability, many head Robotaxi can achieve, even in certain aspects, the latter may do better. However, only a few investors and car technology media (To b), Robotaxi’s car, the TO C class is basically not experienced, so there is no way to do it.
Previously, many Robotaxi’s cars were unwilling to give others test, but according to my colleague Sun Li, Teacher Sun Li, after Huawei “show muscle”, Robotaxi’s attitude towards the test problem will become more open.
However, the short board of Robotaxi is that the supply chain integrates is insufficient, and thus the price of the vehicle is extremely high, which limits the expansion of the team size, which makes it difficult to get a large-scale data, so it is difficult to break through the geographic fence. In contrast, Huawei is through the former load bunter, as long as it can sell 10,000 vehicles with partners in a year, there is a chance to “crush” on the data scale includes Waymo. All Robotaxi companies.
Robotaxi is of course not “sitting in”.
Before Waymo, CEO John Krafcik, marking the direct L4 route encountered a major setback, next, Waymo may enter the domain, steering progressive route through acquisition or investment. Baidu focused on doing L4 and AVP in the past few years, did not do L2, now it seems to be a mistake, of course, by working with Geely, Baidu has also begun to turn to progressive routes. The drip is also directly L4, but through the unified custom car with BYD, even in person with the car, the drip is also very likely to turn to progressive route, first equipped with the L2 kit on the front load branch, and then gradually to L4 Iterate.
In the initial creation, Momenta was in progressive route from the beginning; 4 founders were from Waymo’s soft boat, I was originally intended to do L4 minibus, and recently launched “L3 bus”, actually turning Gradual route. Next, there may be more companies that do direct L4 will also consider progressive routes.
Different from the progressive route of the giants can land through “own car”, the progressive bus from the startups often needs to land in the vendor (usually tier 1), but how to get Tier 2, Tier 3, ensure that key components can pass the car and control the cost, which is a great challenge for these startups; in a sense, for these start-ups, do L2 is difficult than doing L4.
Perhaps, for these start-ups, a better idea is: in cooperation with car companies, start-ups only technical solutions, the supply chain integration is given to the car, or with the supply chain Traditional Tier 1 of integration capabilities, only earn money for technical programs, handle the “bitterness” of the integrated supply chain to the latter. three. “Huawei INSIDE” becomes “Huawei Outside”, To B and TO C boundaries blur
In the second half of last year, a friend of a Robotaxi made a point of view: for technology companies, To b and TO C boundaries began to blur. For example, Momenta has a company, the Robotaxi business is to c, but the L2 business is to b; such as Tucson’s automatic driving truck business, if it is to do a supplier, to b, if the self-built network is directly In the face of end customers, it may be to b or to C.
After the start of the laser radar, there are several laser radar manufacturers even considering that when the timing is mature, advertise in the TO C is “education”, telling them to laser radar. Where is the car that does not load the laser radar, and the laser radar of my home is different from the laser radar of others. After “my laser radar is the most cattle” occupying the C-terminal user, it is great to get the difficulty of the B-terminal customers.
Before the Shanghai Auto Show, it was originally a new car Alpha S, but it was the Huawei of “Party B” under the spotlight. Li Xingyu, Vice President, Skyline Strategic Planning and Market Development, mentioned in the recent “Top Ten Trends in the Times” in the Times, “Brand Propagation of the Extreme Fox Alpha Step, as the Huawei” Chong in front “, put it in front of the supplier In the past, this is very rare, “” The end-to-end intelligent product is also 2C in nature. “
In the words of investors Wang Yuquan, in this wave of PR, “Huawei INSIDE” directly became “Huawei Outside”. Considering Huawei follow-up will also help car companies, “Huawei Outside” is not a joke.
At the time of the media group visit on April 19, Huawei Car BU President Wang Jun said: Our smart car parts business essentially a 2B business, but we do automated driving, smart cockpit, thermal management and other products indeed follow The final consumer’s experience is closely related, so our business is both 2B and 2C.
Li Xingyu commented in the article: “At the end of last year, Huawei’s smart car solution bung into the consumer business department, from 2B to 2C, does not mean Huawei to make a car, but from the side show Huawei to intelligent solution The essential understanding: It is necessary to face the user, design the product, create value. From the technical point of view, do the end-to-end intelligent solution, from 2b to 2C is also necessary, data closed loop inevitable requirements. “
For the problem of Huawei to make a car, Li Xingyu also made comments in the text: “This is more like a pseudo-proposition. When you have become Wintel, do you care if you do a PC machine?”
four. Huawei has no external imagination, “strong”
For many car companies, there is no reason to cooperate with Huawei, the usual explanation is “Huawei is very strong”; in the article “I know Automatic Driving” April 24, a giant Tier 1 China Automatic Driving Director is accepting an interview I also mentioned “Huawei is too strong”.
The author also thought that “Huawei’s scientific giants must be very strong in front of partners”, but from the author recently touched the feelings of Huawei and its partners, Huawei is actually not strong.
For example, although Huawei can provide automatic hot-driven full stack, but MDC’s box with laser radar, millimeter wave radar does not bundle, customers can take Huayi’s box with other companies’ laser radar and millimeters.
For example, although Huawei is very powerful, in the face of a weak customer, their posture is very low, they are willing to do things of British Weida, “can’t see”, serve these small customers.
Compared with the passenger car, the unmanned car is a “very small track”, but the MDC platform is the first in the unmanned mine car scene. A unmanned mine car company, a number of employees, told the author, they have found the British Weida and De Siwei, but they have been rejected. “Our demand is too small, the other party can’t see”, But when you find Huawei, the other party has invested many manpower to support them. Although the previous Huawei did not even heard the names of their company.
Two years, Huawei has provided more than 40 employees to provide support for companies that doing unmanned vehicles. Huawei also admitted that your mine car company is not big, but you gave us a lot of forward-looking thinking.
In fact, in each sub-scene scenario in the automatic driving, Huawei’s resources in “the most inconspicuous” mine car scene are the most.
When Huawei has made a communication business, many resources are often invested for the international giants (including Africa, including US rural), which is also brought into their automatic. Driving business.
Therefore, Huawei’s “powerful” is not necessarily equivalent to “strong”.
As for those vehicles that must master automatic driving technologies, they don’t cooperate with Huawei. Is it really because “Huawei is very strong”? If Huawei is “very weak”, these companies will give up self-research, use Huawei’s full stacking program? Regardless of the new forces or old forces, many companies are not willing to cooperate with Huawei as a “Huawei’s affirmation car”, but the writer’s thinking logic is like this:
Huawei has already taken his business reputation three times, saying that he “does not make car”, but the car enterprises have to refer to the “Huawei does not do mobile phones”, “Huawei will definitely make a car”, do not dare to use his parts, this, Huawei produced a lot of parts to sell, then he can only “forced” to achieve “internal cycle” by personally make a car, so do you have a strong competitor?
On the contrary, if the car company is believed to believe in Huawei “Do not make a car”, we can use Huawei’s parts, let Huawei found that the total profit of selling parts is better than the total profit of the car “,” Then why do he want to make a car?
Fives. Advanced automatic driving begins to work toward “liberation of eyes”
The L2-level automatic driving of the SAE standard is neither liberated, and does not liberate the eyes. Then they will not intervene when they need to take the manual takeover, so at this stage, the so-called “L3 automatic driving” most emphasizes “only liberation of both hands, do not liberate the eyes”, this is not essentially different from L2.
In September last year, Lucid said in the first quantitative car Lucid Air, which plans to provide “liberation of both hands, liberation of both eyes” within three years, compared to those “liberation, but not liberation The so-called “L3” of the eyes “, it is obviously a more daring idea. If it can be realized, it will be a quality flying.
Recently, according to the “42 garage”, D80 + and D130 + also “D80 + and D130 +” of Daxin car launched “belong to the automatic driving of the L3, allowing the driver to get off, and have a short break during the system operation.” However, as for vehicles carrying D80 + and D130 + in automatic driving, the accident responsibility that occurs when the driver just “looks off” should be borne, and there is not mentioned in the report.
In theory, only in the car enterprise or automatic driving plan supplier promises “you can rest assured to ‘get off’, responsible,” consumers dare to “look off”; but from the perspective of practice, no matter the car Still the planner, I don’t dare to take this responsibility easily. In addition, Next, DMS will become the basic standard equipped with advanced autopilot system for a vehicle, if the manufacturer allows the driver to liberate his eyes, that DMS should not pretend? If installed, how does the function definition?
Of course, in practice, Daxin car and cooperative car companies will find that this setting is too optimistic, thereby restricting the driver “deprotection” in the product manual.
six. There must be “soft and hard integration” ability, and there must be a “hardware and software decoupling” posture
Huawei was originally a hardware company to do, but now there are 1,200 people actually doing the autopilot algorithm, accounted for 60% of its autopilot team, which is surprising. Based on the slimming 310 chip and the MDC platform, Huawei also self-developed the operating system, a perceived algorithm and other software, and also self-developed a sensor such as laser radar, sensor, and camera, and constructs its true hardware integration capabilities.
From the capability system, the Dajiang (including Dajiang car and Livox) with Huawei is like, and are equipped with self-developed automatic driving computing platforms, laser radar, millimeter-wave radar, the ability of software algorithms.
In terms of hardware and software integration, Horizon is also of concern. Its founder and co-founder primarily algorithm background, but made a hardware-based chip company; and in the actual conduct of business, they make full use of their own advantages in the algorithm, algorithm development to help car firms. Actually provide hardware and software integration solutions.
AI computing hardware and software have high demands synergistic, efficacy greater good software algorithms, hardware allows play; and vice versa. Therefore, compared to only software or only hardware companies, those companies with hardware and software integration capabilities will be more competitive.
For suppliers such as Huawei and Dashi, hard integration capabilities are the air of the full stack of solutions.
Standing angle car prices compared are from different manufacturers combination regimen hardware, software, tools chain, having a full stack Program “extreme end performance optimization, efficient responsibility clear interface, internet smooth evolution” unique advantages, use up too convenient; moreover, full-stack solutions make their own car can be achieved earlier autopilot – not the courage car prices have put a lot of resources to research since the autopilot algorithm, even if put, there is no full Grasp can succeed. These car prices, the full stack plan is in fact the best option – to do technology companies “foundry”, than sales continue to decline, even sustained losses, bankruptcy.
According to media public reports, the founder of the Skyline founder Bao Junwei made a suggestion for chip and laser radar during Baidu, but did not be adopted at the time. Since then, Baidu has invested Velodyne and Herpet in the field of laser radar, and a laser radar of a cooperative development facing a front-loaded mass production. Next, Baidu may also design the automatic driving chip. However, it should be noted that Huawei, Daxin and Skyline, although it has a soft hard integration capability, but is open in business model, namely software and hardware decoupling, hardware with hardware decoupling. For example, Huawei and the calculation platform of China can be equipped with laser radar and algorithm of others; and the horizon is even more than the “not bundled” strategy to grab many customers from Mobileye.
Of course, Huawei also emphasized that only those models with Huawei automatic driving full stack solution can only hit “huwei inside”, which means that even Huawei provides a software and hardware decoupling program, However, those self-product power is not strong, I hope to enhance the brand value of brand value through “hi”, it will still be more integrated with its full stack of programs when working with Huawei.
In general, “hard integration” capacity must have, which is the key to the competitiveness of technology and products; at the same time, “hardware and software decoupling” must have, this is to win partner trust Foundation.
Seven. The two trends in force stack and force reduction are in parallel.
In January of this year, it was a prelude to automatically driving “Medical Army Competition”, but no matter how powerful, it is only a transition plan, etc. It is output, it may still be subtracted.
Finance and Easter “” “Huawei’s first truth” in the first truth of “Huawei” said: “In the case of ensuring overall functional force, Huawei provides calculation modules such as memory particles and chip particles. I have already thought that it is a low cost module, which is also considered for the car. “
According to the engineers of Dawang, “With the excellent algorithm of Daxin, even if the D130 + system of the L3 level, the strength demand will not go to the exaggerated hundred TOPS.” This means that subsequent, some software algorithms have accumulated a deep company, will not be passively involved in the “Medical Army Competition”.
Eight. Is the laser radar really just “redundancy”?
In the past few years, the automatic driving circle has a general statement of the sensor program is “laser radar is the main sensor, the camera is auxiliary sensor”, but in the past two years, more and more people have begun to say “to do the main sensor with the camera” , Laser radar as redundant sensors. However, “redundancy” is questioned.
At the end of November 2019, Furik CEO Zhang Lin explained this when talking about the word “main sensor / auxiliary sensor” in writing: In the field controller era, it is basically a fusion sensation algorithm, we will not talk Who is the primary sensor, who is a auxiliary sensor.
In July last year, the author asked the other party when he communicated this topic with a laser radar company CEO: “How to define who is ‘main sensor’?”
At that time, the author quoted the interpretation of a new energy vehicle enterprise automatic driving person in charge: “When the identification results of the two sensors are inconsistent, the decision system decided to pay attention to who’s opinion, who is the main sensor.” For this, this laser The response of the radar company CEO is: “If the identification results of the laser radar and the camera are inconsistent, you have to listen to the camera, then remove the laser radar, if the two are inconsistent, listen to the laser radar, then The meaning of the camera called ‘main sensor’ is not big – it is a bit like saying ‘Our home is the main thing, but when the opinion is inconsistent, my wife said it’. “
This logic is still very persuasive.
This time, Huawei ADS person in charge said: “I think the sensor doesn’t have Redundancy (redundancy), this is a crap. After two years, we abandoned, now we all They are all fused technologies. “In addition to the meaning, if the perception is the front convergence plan, who is the primary sensor, who is redundant, it is meaningless.
Recently, “No. 42 garage” Bai Yang also 怼 “Laser radar is a redundant sensor”: “‘Redundancy’ means, even if it is not, you can get it. There is an alternative to hunch; and at this stage, the laser radar is not only redundant, but also the main force of some of the functionality of some scenes, this is clearly the live room of the spare tire. “
Usually, the automatic driving company is talking about “laser radar to do redundant”, it is implied that “my visual algorithm is already very busy”, but this may be dug a pit. The perceived algorithm engineer of a new power of the car said: “The first principle of Mask is that it is no problem to stand in the perspective after 10 years, but the problem is that the current visual algorithm has no revolutionary breakthrough, he The technical route is to be perceived with the blending route. “
Next, companies will continue to work hard to improve visual algorithm, but as long as you haven’t completely get rid of the laser radar, then who is the primary sensor, who is redundant, more like a text game. nine. The decision algorithm is not good, and the number of sensors is useless.
In a media group visit on April 18, Su Yu’s mobile phone mentioned in response to the Mobileye mentioned radar, Lidar as a subsystem, and said pure visual as a subsystem, independent test, “said” “truly determines the accommodation rate It is not only a sense of perception system, and the relationship with the regulation is very large. Even the weights occupied by the regulation are more than the perceived system. “” Most of those difficult cases, you add 80 times the sensor to the sensor Can’t. “
By the sensor stack, it does not fundamentally solve the problem. The author recognizes this view of Su.
In August last year, Dr. Guo Jiyi, the author and the GAC Research Institute, which is more difficult to understand the decision-making algorithm, Dr. Guo Jiyi’s answer is: “The perception is an easy standardization, and the decision-making algorithm is easy to make differentiated, we Evaluating a cow is trying to force, not to see his vision, but look at his thinking cows. “
In the L2 stage, because people are the main body of the driving responsibility, all kinds of automatic driving research and development center are in a perceive algorithm, so whoever is perceived, whoever is “more bursts”; but arrived at the advanced automatic driving stage, The main body of driving is gradually transferred to the system. At this time, it is really difficult to behavioral prediction, path planning and control, namely decision algorithm and control algorithm. Simply put, 20 myopia glasses can’t make up for the brain is not smart enough, and the experience is not rich enough.
The algorithm engineer of a new power of the car said: “Do not understand the whole technical link manager, you will think that the current perception is limit. Perceived is of course a problem, but not the main reason. We are so hard to go to the car, the camera, Microphone, speaker, incurred logic, plus AI brain, but to let the perceived integration cap of the car as close as possible to the lower limit of the perceived ability, but from the perspective of the human perspective, to achieve function, unlimited pursuit of perceived data Precision, will it fall into an algorithm trap? “
ten. Low line digital laser radar market will be eaten
In recent time, 4D millimeter Prada concept is particularly fire, of which 4D millimeter wave radar is likely to replace laser radar is one of the hot topics. The author’s point of view is: In some scenarios, 4D millimeter wave radar can replace the low-line digital laser radar (such as 32 lines, 16 lines), but it is impossible to replace high-line laser radar.
Huawei car BU president Wang Jun said in an interview with the media: “We have mentioned the 4D millimeter wavelet when it is required, and the performance must be able to replace the low-line laser radar, so force the laser radar to the high line, thus forming Bright circulation, millimeter radar and laser radar technology are desperately walking, and then cost the cost. “
As the 4D millimeter wavelet market is increasing, the market in low-line laser radar will shrink. Of course, the laser radar is originally developed from the low number of high lines, even if there is no 4D millimeter wave radar, the low-line laser radar will also be step by step by a high-line laser radar. This belongs to the “self-revolution” of laser radar manufacturers.
In the future, the trend of laser radar manufacturers will be more obvious to the high line of digital products, which is a great harmful to high-line laser radar.
eleven. The map business must speed up the high-precision map collection of urban roads.
At present, when the major maps are talking about how many roads have covered their high-profile maps, the data taken out is about high-speed roads, and few people talk about urban road data. Because, everyone is determined, the high-level automatic driving ability will take the lead in the high-speed scene, so put the center of gravity on the high speed. However, now they will find that the situation has changed.
Huawei automatic driving, no high speed from the beginning, but first focusing on the commuting issues of urban residents. Americans driving more time, so Tesla’s automatic driving ability is started from high-speed scenes; in contrast, the Chinese driving is on the road road, so Huawei’s technology The choice of landing scenes is undoubtedly more appropriate.
Huawei has a few domestic high-precision map vendors in 2019, but at that time, these manufacturers did not collect maps of urban roads. Therefore, Huawei applied for high-precision map collection qualifications, and decided to collect high-precision maps of urban roads. Next, the vehicles equipped with Huawei’s high-precision map gradually, and quickly expand the high-precision map coverage, quickly update the high-precision map, and the high-precision map of the map will not keep up with the coverage of urban roads. Very passive.
In addition, the update of high-precision maps, if you have enough car rules, you can realize automatic updates, that is, the vehicle will automatically discover the difference between the actual road conditions, and then trigger the data, and guide the capture of the car. It will greatly improve the efficiency; if the number of cars is too small, the difficulty of “automatic discovery” by the vehicle will be large, this needs to be planned to plan, such as one month update – maybe all the roads do not divide all the roads Re-sweep, resources and energy waste is extremely serious.
It can be said that if Many will not provide automatic driving overall solutions, or work with the depth of the car, it will be very passive in the update of high-precision maps. twelve. “Little Business” in the big scene and the “big business” in the small scene
During the auto show, I met a unmanned mine car company CTO. The other person told me a very interesting information: “Terminal logistics is nearly 100 billion yuan market, but the market is scattered, there is a strong region, you are doing in this city Ok, go to another city, it may not be able to do well, and, the amount of each community is very small, and the operating cost is difficult to amateur; in contrast, the open coal mine transport business, the market volume is only 30 billion, but Each mine is very large, as long as you take a deposit of a mine, you can earn a few billions a year. “
This information is very valuable. When the automatic driving company is looking at the application scenario, it is not possible to focus “how big this track is”, and it is necessary to assess how much difficulty is the difficulty of winning a certain market share. In summary: If the direction is not matched with his own capacity system, in the big scene, you may only do a small business; if the direction is selected, in the small scene, you may also make a big business.
thirteen. Can’t simply put your smart car ratio as your smartphone
In the past few years, we have found that there are many inner people, especially media large v, always like smartphones to do the ratio when analyzing the prospects of the smart car industry, this kind is actually not reliable.
On April 18th, “Late Late POST” reporter has given a question from Huawei Car BU President Wang Jun: “Huawei has across the mobile industry to the automotive industry, there is no experience that can be replicated, but the actual difference is very large Place? “Wang Jun’s answer is:” The mobile phone is the consumer level, with the car home, the industry is industrial grade, the car grade. We have arrived in this industry, found that the requirements are completely different, the mobile phone except the battery, for personal safety The impact is relatively small, but many aspects of smart cars are closely related to life. “
Huawei Intelligent Automobile Solutions BU Smart Cockup Product Department General Manager Wang Qingwen added: “From the perspective of end users, smart cars are very close to smartphones, but from engineers, product development perspectives, this is two very different Things, because its life cycle is different, and the life cycle will bring software development, software management and hardware management are completely different – mobile phones may change once for two years, so software version does not need to consider hardware; while the car life cycle is usually More than ten years, if the software is upgraded, any version inside the car needs to match the hardware matching of stock, this workload is huge. “
Wang Jun said: “How to do the life cycle management of the software, we are still exploiting, especially in the cockpit field involves frequent upgrades, life cycle management is very complicated, there are too many versions to maintain. So, software definition car, OTA It’s really not simply, it is necessary to consume a lot of manpower, material, and the process guarantees that this thing can really land and implement. “
fourteen. Those companies that “lack aura” may need to be revalued
Previously, the automatic driving also stayed in the test stage. Everyone is in the strength of the technology, and the importance of the project capability is highlighted. Simply put, the technology of the Internet, AI background is stronger, but the technical ability, the traditional Tier 1 and car enterprises may have “rolling” advantages.
A few months ago, a CEO and CTO were from the Delta Tier 1 domain controller company’s friends said: “Our boss feels very wronged, we have been profitable, it is better to lose money for a valuation, even life and death L4 automatic driving startup company high? “The author joked:” Because, in the eyes of investors, your company has no ‘technology sense’, lack of aura. “
Automatic driving company founder, or even CTO is usually said to be “segment”, even “self-contained red attribute”, and the founders of many traditional Tier 1 is often “engineers of honestly”. ” In extreme cases, in extreme cases, the company’s investment and finance department and market department staff are also unknown, and when they see investors and media, “I don’t know what I do.” This leads that the value of these companies may be seriously underestimated.
However, when automatic driving from the story of the story, the value of these companies may be further excavated, and even the valuation may rise sharply.
fifteen. “High-end” may become an disadvantage of Apple
Xiaopeng P5 release and the cooperation with General Wuling, and the high-level automatic driving in the price “civilians”, this thing is very significant.
The earliest production car MODEL X and MODEL S, then the later Audi A8, the price is basically more than RMB 800,000; Cadillac CT is more than 64 million; Xiaopeng P735 million; carrying laser radar Xiaopeng P5, is expected to be around 200,000; Next, the quantification of the quantitative car with General Wuling Baojun, the price may be controlled around 150,000, even lower; Xiaomi’s first car should also be at this price.
Previously, when the price of the chip and sensor was more expensive, the automatic driving system can only take the lead in the user’s high-end models that are not sensitive to price, but with the intensification of competition and the decline in supply chain, more and more The medium and low-end models will “fit” good automatic driving chip and laser radar. He Xiaopeng has mentioned that “The more high-end customers are more interested in automatic driving, the more high-end customers, the luxurious and dynamic demand for the car”, the early days of the industry, this view is basically established, But we can’t understand “” Medium and low-end users are not interested in driving “, they just don’t pay enough financial resources to pay for their interests, so I have” pretending to be not interested. ” With the rapid decline in the cost of automatic driving kits, those who originally docked automatically driving “no interest”, soon become “interested”.
What is not the same as a smartphone is how the automatic driving system is largely depends on how big is the size of the team. Therefore, standing in the perspective of the vehicle enterprise, in the hardware configuration, the user’s optional rate is almost the premise The higher the models, the higher the sales, the greater the contribution to the advancement of the company’s automatic driving algorithm. After all, the competition of each company’s automatic driving ability is in the short term.
Therefore, if the vehicle enterprises feel that the automatic driving algorithm is the first major event, in the case of limited resources, the automatic driving system should be taken in the model that is most sales, easy to become explosive models, not in the “highest end”. On the model – if you are equipped with the automatic driving system, although the high end, only a few thousand vehicles have been sold, the value of accumulation data is limited. If it is possible to make the automatic driving as an ideal plan, it is of course best.
From this perspective, Apple does not necessarily enter the first echelon, despite talented, but automatic driving capabilities may not enter the first echelon. Because Apple’s adjustment is determined, their products must be “ultra-high-end”, we assume that Apple is standard for automatic driving kit on the car of RMB 600,000, and the buren or millet is a car in RMB 150,000. Said the same standard or a little automatic driving kit, then who is more favorable in the scale, who is the algorithm to progress faster?
Of course, the big bodies or millet must also consider how T-box on the car, how to return data, and how to deal with data – no processing ability, the data is more useless; in addition, it is equipped with automatic The EE architecture of the first car of the driving system should be a centralized architecture, which can be supported in the next car OTA.