Image source @ Visual China
Text | Car smart driving
Previously, as one of the first batch of pilot enterprises for the commercialization of autonomous driving in China, Baidu officially launched the autonomous driving travel service platform "Radish Run" in August last year.
This is not only the first landing of Baidu's autonomous driving commercialization strategy, but also the first autonomous driving travel brand in China to try to provide paid services, which is of great significance to the entire travel track. After a little half a year of operation, as of the fourth quarter of 2021, the number of manned orders for radish run has reached 213,000, and it has opened up Shenzhen, Shanxi Yangquan and other cities.
At the same time as Baidu, Xiaoma Zhixing, which opened the commercial fee operation pilot, also completed the first delivery of the D round of financing in February this year, and the valuation of the company also rose by 65% to 8.5 billion US dollars.
As a result, citic Securities, a well-known research institute, boldly predicts that autonomous driving has a great possibility and will become the next potential track with a trillion market width after new energy vehicles.
However, in addition to flowers, applause and countless praises, such doubts continue to linger in the minds of many people - has not really crossed the L3 threshold of autonomous driving, has really reached the stage where it can be commercialized on a large scale?
L3 is not yet widespread, and autonomous driving is still young
According to SAE's Society of Automotive Engineers grading standards, autonomous driving technology is divided into 6 levels from L0 to L5. From the L0, which is simply equipped with automatic emergency braking and various hazard warning functions, to the L1 and L2 that initially provide drivers with control functions such as steering, braking, and acceleration, they are strictly speaking only "intelligent driving assistance".
As a watershed that distinguishes automatic driving and driving assistance, L3-level automation technology can only achieve automatic driving under the premise of meeting specific conditions. And the vehicle is still inseparable from the driver's cooperation, in the driving process, it is still based on manual interference.
Once you cross the threshold of L3 and enter L4, although there are still demand restrictions on the driving environment, you have been freed from the dependence on the driver, and even from this stage, the pedals and steering wheel can be removed from the vehicle. If you go further to L5, there is no restriction on automatic driving, and it belongs to an automotive robot that can be used in any scene with confidence and has high intelligence and high interaction capabilities.
Baidu's regional unmanned taxi can be seen as a specific manifestation of L4 autonomous driving technology. It is only based on multiple considerations in terms of safety and public opinion, and at this stage, it is still equipped with a safety driver to deal with emergencies, so strictly speaking, it is more like a technical application in the late L3 stage, also known as L3+.
Looking at the current new energy vehicles in the world, there are very few corners that can realize the L3 level automatic driving function. Even if the industry generally believes that 2022 will become the first year of L3 autonomous driving mass production, at least at the current stage, most of them are still in the late stage of L2.5 and L2.9, and many car companies cannot even achieve L1.
Mercedes-Benz, which is expected to mass-produce vehicles equipped with the L3 level autonomous driving system DrivePilotL3 in 2022, became the world's first un-R157 certified automaker last year. International traditional car companies such as BMW and Hyundai are also at the critical point of about to mass-produce L3-class models, excluding the time spent on testing and obtaining qualifications, and there may be no short way to go to truly achieve mass production of L3s.
In contrast, the new energy vehicle companies that have emerged in China in recent years, relying on the accumulation of their own technology, most of them have relatively perfect L3-level technology, and have already aimed at the large-scale production of L4-level high-end automatic driving. In fact, at least in the next two years, the companies that can achieve L3 mass production and even L4 landing are still only a small number.
Coupled with the current shortage of chips in the downstream of the supply chain, as well as the epidemic environment, the global upstream foundries are also generally insufficient production capacity, and it is difficult to develop automatic driving that is extremely demanding on software and hardware. In a short period of time, L2+ and L3+ may be the general level of the autonomous driving market.
Moreover, in addition to the limitations of production capacity and research and development capabilities, the relative lag of supporting equipment and technical systems will also hinder the large-scale mass production of automatic driving. For example, lidar, main control computer, sensors, cameras, etc. Automatic driving basic performance structure, at this stage, there are also various doubts, similar to whether lidar will be IQ tax, whether the camera invades the privacy of car owners, whether automatic driving is safe, etc., are constantly increasing the panic of consumer groups about unfamiliar technologies.
Even without considering the technical aspects, although Baidu's regional driverless taxis have proved their commercial viability, they are still limited to parts of first-tier cities. It is a long process to promote it to the whole city, or to gradually sink from a first-tier city, and it is inseparable from the long-term cultivation of the entire social environment.
In the view of car intelligent driving, in the field of passenger cars, autonomous driving is still in the early stages of a very young age. Just like the sun at five o'clock in the morning, although the potential is extraordinary, it is still a lot of time away from the glory of the sun, and it is too early to want large-scale commercialization.
Of course, in addition to passenger cars, autonomous driving technology also has a very broad market segment. Will the current level of technology be able to "advance" the results in other areas?
Can the market segment carry large-scale commercial use?
At this stage when the L2-level automatic driving function has just entered the market popularization period, according to the speed requirements of the automatic driving scenario, it is estimated that the difference between low speed and high speed can be used to classify the relevant commercial scenarios.
In the low-speed field, most of them are the distribution of goods in closed parks, communities, ports and other environments, as well as the application of fixed scenarios such as unmanned sanitation, and the functional requirements are relatively professional. In the high-speed category, most of them are based on daily cars, unmanned taxis, intelligent logistics trucks, etc., and the physical width of the scene is broader and the needs are more complex.
Among them, logistics distribution in a closed environment is mainly reflected in unmanned delivery robots, unmanned express delivery vehicles, drone takeaways and so on. These autonomous driving segments, which are promoted by leading technology companies such as JD.com and Alibaba, have been put into practical use in the initial outbreak of the epidemic and have achieved excellent market feedback.
In the current segment of the market involved in autonomous driving companies, the mainstream development direction that is recognized as the most potential is unmanned sanitation, unmanned taxis, intelligent logistics trucks and automatic parking.
With the continuous increase in the growth rate of urbanization in China and the serious aging of sanitation workers, the market demand for unmanned sanitation vehicles is expanding. Based on the needs of health safety and urban appearance construction, local governments have also introduced policies to support the commercialization of unmanned sanitation.
Nowadays, it has attracted countless enterprises to enter the field of unmanned sanitation. At present, there are enterprises similar to Xiantu Intelligence, Kuwa Robot, etc., which cut into the track with L4 level automatic driving technology; Shenlan Technology, Xidi Intelligent Driving, etc., which are committed to the research and development and application of artificial intelligence; Gaoxian Robot, Scitech Intelligence, etc., technology-driven enterprises based on service robot products. Looking at the sky eye, there are already 1424 companies registered in the case, including many well-known companies.
Although there is a difference in speed, there is not much difference between the two in terms of performance and hash rate requirements. All that needs to be faced is a more complex dynamic working environment, and high perception, high AI, and high computing power are also indispensable keys.
If there must be a distinction, it is the high-speed autonomous driving segment with a wider application area and longer threads, so as to carry the carrier of technology applications, the cost and maintenance cost are more expensive, and the safety risks involved are more obvious. Moreover, in addition to the higher return on investment, high-speed automatic driving has also become a higher entry threshold, and the players in the current track are still dominated by major car companies and travel companies.
In the field of unmanned taxis, in addition to Baidu and Xiaoma Zhixing, which is currently leading the industry, Wenyuan Zhixing, who has already conducted the safety test of unmanned taxis in Guangzhou in 2019, is also a player that cannot be ignored. And not only passenger cars, but also public vehicles that are widely used in urban scenes, such as unmanned minibuses and unmanned light rails, are also battlefields that related enterprises cannot avoid.
It is also more unmanned parking for the ToC market, mainly used to solve the demand pain points in the daily use of family cars, and has great market potential. At present, Wenyuan Zhixing, Yutong, Renault Nissan, Xiaomi, GAC and other enterprises have already made practical landing attempts at the relevant level, especially after the gradual improvement of the IOT Internet of Things ecological environment, the automatic parking function with obvious low-speed characteristics is becoming the resident function of intelligent cars.
The intelligent logistics truck to solve the demand for ToB, because the country has not yet opened the unmanned heavy truck test on the main road, the current unmanned truck is mainly used in lightweight trucks, and is still in the port, logistics park and other exclusive scenarios. Unmanned trucks between low speed and high speed mainly solve the problems of driver safety and labor costs in long- and short-distance transportation.
However, although in theory, L4-level automatic driving technology is the core foundation of intelligent logistics trucks, under the influence of policies, market environments and other aspects, L2+ level automatic driving has begun to land. Beidou navigation, which has been applied to the field of automotive assisted automatic driving, may be the best proof.
There is no doubt that the market segments related to autonomous driving have market potential that cannot be ignored, and the L2 level autonomous driving technology in the popularization stage seems to have met the commercial premise of some industries. However, Chezhi Driving believes that regardless of the gradual approaching of L3-level mass production, the upcoming industrial changes, from the most basic business logic, now is not the best time for large-scale commercial use of autonomous driving.
Revenue is under pressure, and safety hazards cannot be ignored
The commercialization of any technology is inseparable from the most basic business logic, that is, the discussion of cost and revenue. Under the huge market potential, autonomous driving related companies may not have to worry too much about revenue data, but how much it costs is a difficult problem to ignore.
As a type of technology that can cross the industrial structure, autonomous driving can be imagined to be difficult to develop. Baidu, which bet on the future of the entire company, has insisted on it from 2017 to the present, at the cost of countless funds, time, and countless missed outlets, in exchange for the Apollo intelligent system, and the possibility of becoming the first enterprise in the entire industry to realize the commercialization of L4 level autonomous driving technology.
Even so, due to the lack of basic links, Baidu unmanned taxis only show the technical level of L3+. Perhaps until 2025, equipped with Qualcomm's next "computing ceiling" high-performance chip and highly customized intelligent cockpit platform Jidu automobile mass production, in order to truly realize the technical power accumulated for many years.
Most of the companies in the industry that are trapped, and may be trapped for a long time, can not get rid of human interference L2 level autonomous driving technology, just the research and development pressure of autonomous driving is enough to seriously squeeze the profit margins of the entire enterprise. For example, Baidu's research and development expenses in the third quarter of 2021 exceeded 6.2 billion yuan, which is almost astronomical for small and medium-sized enterprises.
In other words, in the market is generally in the L2 stage, the profitability of autonomous driving related industries, not a small part of it is walking on the edge of loss of small profits, if there is not enough capital and technical support, most companies may have to stop here. If you really want to make a profit, you may have to wait for the L3 level autonomous driving to complete the cultivation of the market and increase the overall revenue in the L4 market.
In addition, on the basis of existing technology, the core issue of achieving automatic driving or not is traffic safety. However, in fact, not only in the test link, the autonomous vehicles maintained a not low accident rate, but also recently, Tesla, Weilai, Ideal, Xiaopeng and other intelligent models have all had traffic accidents caused by automatic driving.
Driving safety is not a cold number on the table, although autonomous driving is not equivalent to zero accidents that are 100% safe, but at least at this stage, the frequency of accidents does not meet the standards that automatic driving really should have. The most basic security cannot be guaranteed, and large-scale commercialization naturally cannot be said.
In addition, autonomous driving requires a lot of data collection and scene feedback, which is inseparable from the whole process of networking. Whether it is a road signal source or vehicle WIFI, it is likely to become the target of network hackers. What it means for an automatic system to get out of control in a driving scenario is clear to everyone.
Based on the current market environment, the auto intelligent driving editorial team judged that although autonomous driving has begun to prove the prospects and feasibility of its own commercialization at this stage, if you want large-scale commercialization, you may need to wait for the promotion of L4 level automatic driving technology and the improvement of network information security construction.
Before that, let's do a good job of cultivating the market.