Recently, four "bigwigs" in the smart car industry have launched a heated debate in the air around the core component of lidar.
This protracted discussion stems from the fact that in mid April, Jidu automobile first exposed the company's initiative of double lidar automatic driving technology scheme with headlights and front cover.
Ideal car founder Li Xiang posted a microblog to question this plan, which led to Jidu car CEO Xia Yiping's personal explanation. Subsequently, he Xiaopeng, chairman of Xiaopeng automobile, and Shen Hui, founder of Weima automobile, also successively focused on "how many lidars should be installed and where?" And other issues.
The heated debate of the four leaders of automobile enterprises has raised the heat and attention of the lidar sector. In fact, since the second half of last year, many car companies have announced that they have equipped their smart cars with lidar, and even launched an "arms race" for the specific number of vehicles. Will the application of lidar significantly accelerate the improvement of vehicle automatic driving level? When will lidar usher in the first year of large-scale boarding?
Ma Tianyi, chief analyst of the communication industry of Minsheng Securities Research Institute, said in the "V interview" with E company that at present, the technical path of lidar in the industry has not been unified, and China's relevant industrial chains still need to further clarify the division of labor. Facing deep integration, the cost of lidar itself has not been completely reduced. On the whole, it is very difficult for lidar to realize large-scale boarding in 2022. This year can not be called the first year of batch boarding of lidar.
pure visual route vs multisensor fusion sensing
In recent years, with the continuous improvement of automatic driving technology, the industry's requirements for its perception layer have gradually increased. At present, automatic driving is divided into three levels: perception, decision-making and execution. Among them, the main sensors of the sensing layer include camera, millimeter wave radar, ultrasonic radar, lidar and infrared sensor.
Up to now, there are still route disputes about the configuration of automatic driving sensors in the industry. One is the "vision school" led by Tesla, and the other is the "multi-sensor fusion school" dominated by new forces of car making, that is, multi-sensor such as lidar and millimeter wave radar are equipped on smart cars.
Since 2021, more and more automobile enterprises have begun to turn to the "multi-sensor fusion school" with lidar as the main body. Weilai, Xiaopeng, ideal, Weima, Nezha, Zhiji automobile, GAC AIAN, Jihu, avita and salon automobile all claim to have launched or will launch smart cars equipped with lidar. The number of lidars carried in new cars has also increased from one to three or even four.
\u3000\u3000 "We can compare the perception layer of automatic driving with human perception system. In Musk's cognition, automatic driving vehicle is like a 'child', and its growth needs a certain learning process. It needs continuous machine learning through artificial intelligence to become smarter and smarter; while Chinese car enterprises hope that automatic driving vehicle is a 'Superman', and the application of lidar can help drivers perceive those humans Invisible corners. " Ma Tianyi told the reporter of the securities times · e company that Tesla has a very strong algorithm team, so musk can take the camera as the core perception system of its vehicles and continuously carry out machine learning, but it is difficult for Chinese car enterprises to bear all kinds of risks in the learning process, so they rely more on lidar for perception.
Ma Tianyi explained that the emergence of the two different technology paths is actually related to the technical foundation and later decision-making of each car enterprise. In his opinion, musk may also launch vehicles equipped with lidar in the future, which will be parallel to the technical scheme of pure vision, and most Chinese car enterprises will choose the fusion perception route with lidar as the core.
large scale boarding still needs time
Lidar, known as the "eye of machinery and equipment", is widely regarded as the core sensor of intelligent driving vehicles above L3 level in the industry. However, due to the high cost, lidar did not usher in large-scale application in the automotive field.
In recent years, with the outbreak of the global intelligent automobile industry, the boarding of lidar has increasingly become the general trend. Vehicle enterprises such as Weilai, Xiaopeng, ideal, Weima and Nezha have equipped lidar for their intelligent automobile products. On April 25, when Nissan announced the research and development progress of the new driving assistance technology, it also revealed that the technology integrates the core information of the next generation of high-performance lidar.
Previously, it was widely rumored in the industry that 2022 will be the first year of large-scale boarding of lidar. In this regard, Ma Tianyi believes that it is difficult for smart cars to apply lidar in batches this year.
There are several specific reasons: first, the technical path of lidar has not been unified and determined. Is it mechanical lidar or electronic lidar? There are differences in the industry.
Secondly, even after the route is unified, there are differences in the wavelength of laser light involved in lidar products, which directly affects the cost. Ma Tianyi believes that under the background of certain macroeconomic pressure this year, there will be great pressure for car enterprises to realize the large-scale boarding of lidar.
Finally, although China's intelligent automobile industry is in a state of vigorous development, and many car enterprises are expected to realize "corner overtaking" with the help of the trend of electrification and intelligence, lidar is a part after all. Whether the application of this part will become the core factor driving consumers to buy a car remains to be observed and discussed.
Ma Tianyi believes that car companies publicize that their products are equipped with multiple lidars, which will attract more attention, but at present, the large-scale boarding of lidar will take some time.
lidar cost decline is the general trend
Since this year, covid-19 epidemic, rising raw material prices and chip shortage have become important factors disturbing the development of intelligent vehicles.
Ma Tianyi believes that this round of sharp rise in raw material prices may drive the redistribution of profits in the manufacturing industry. In the short term, the immediate impact has a certain impact on the boarding rhythm of lidar, but in the long term, the development progress of intelligent vehicles will not be impacted.
Although the industrial development trend has been determined, the cost of lidar is still not low. Relevant data show that at present, the price of mechanical lidar is generally high, which is mainly used in test vehicles. The price of 64 wire lidar of velodyne, a lidar company, is US $80000, while that of 16 wire lidar is US $4000. The average price of vehicle mounted semi-solid lidar is about $1000.
Ma Tianyi believes that to drive the cost reduction of lidar, we must first realize large-scale boarding, and secondly, the macro environment of the industry should be good enough.
"In addition to the above two points, car companies also need to think about what they can do in the car and how to build the subsequent business model after the improvement of automatic driving technology and the liberation of drivers' hands". Ma Tianyi believes that if car companies can solve this problem and combine the macro development trend, the cost of lidar will be reduced closer and closer.
Earlier, Huawei had announced plans to reduce the cost of lidar to $200 or even $100 in recent years.
In this regard, Ma Tianyi believes that it is difficult to significantly reduce the cost of lidar in the short term. The specific reasons are divided into several points: first, the cost reduction of lidar requires "volume price matching", and it is impossible to reduce the cost without increasing the scale; Secondly, at present, many upstream enterprises of lidar in China, including manufacturers of light sources, modules and prisms, have just started, and have not been well matched with downstream manufacturers. Further combing of the industrial chain requires a certain time cycle.
However, Ma Tianyi also stressed that in the long run, the decline in the cost of lidar is the general trend. If it can be reduced to about 1000 yuan, it is also reasonable for the whole vehicle to be equipped with more than three lidars.
after clear division of labor, opportunities for domestic manufacturers are significant
It is understood that at present, the major lidar manufacturers worldwide include velodyne, luminar, AEVA, ibeo, Ouster and innoviz. China's Huawei, Dajiang, Hesai technology, Suteng juchuang and Vanjee Technology Co.Ltd(300552) are also the core players in this track.
Among them, the P5 of Xiaopeng car is equipped with the lidar developed by Dajiang; While Nezha s, salon machine Jialong, extreme fox alpha s and Chang'an avita s use Huawei's lidar; The products of auto enterprises such as Gaohe, Aichi and Jidu have applied the lidar of Hesai technology; GAC AIAN, SAIC Zhiji and Weima choose to configure the laser radar of Suteng juchuang.
Ma Tianyi told the reporter of securities times · e company that at present, most domestic models give priority to carrying domestic lidar products, while overseas models choose more overseas lidar, so there is a natural distinction in application.
"Both upstream optical devices and downstream lidar manufacturers, China has certain advantages in various sub fields of the lidar industry, even ahead of foreign countries. At present, the rapid development of domestic intelligent vehicles also provides good development soil for local lidar enterprises," Ma Tianyi said. There is no domestic substitution logic in the lidar field, The advantages of local lidar enterprises are very significant.
According to Ma Tianyi, in 2010, among the top ten optical communication enterprises in the world, China accounted for only one seat; By 2020, six optical communication enterprises in China have entered the top 10 of the list.
According to Ma Tianyi, the most obvious difference between local lidar enterprises and overseas companies is not in the technical level, but in the choice of technical path. However, up to now, everyone's technical route is gradually approaching unity.
However, it is worth mentioning that at present, China's lidar industry chain needs to be improved, and many details are still being built, including lidar terminal equipment manufacturers, chip manufacturers, module manufacturers, etc. the division of labor is not very clear.
Ma Tianyi suggested that manufacturers of lidar terminal equipment still need to focus more on their own links and fields, and the upstream work can be completed by upstream enterprises. It is a more reasonable model for everyone to do a good job in the division of labor and grasp one of the links.
At the same time, he also suggested that it is necessary for lidar manufacturers to invest a lot of energy in adapting to car companies and seize the track in advance.
"It's not easy for car companies to add a new component to the vehicle, especially the lidar is also related to the safety and reliability of autonomous vehicles, which requires more caution". Ma Tianyi believes that although car companies are relatively strong in the industrial chain, lidar companies need to strengthen communication, actively cultivate car companies to adapt and speed up the boarding of lidar.
In addition, in order to avoid the overheated development of the industry and the rush of communication enterprises, Ma Tianyi also suggested that car enterprises appropriately raise the threshold of lidar boarding, which is very important to ensure the safety and reliability of vehicles and is more conducive to the long-term and healthy development of the industry.
