2022 LiDAR is facing the first year of mass production, but it is still far from large-scale mass production| the evolution of car manufacturing

Coverage of the digital economy

Define a transformed China

To truly realize the scale of lidar on the train, in the final analysis, it depends on the technical barriers and the feasibility of technology.

Written by | Tabornet one light-year

Edit | Shikino

Technology and capital are running wild, consumer demand is not reduced, and the development of automatic driving has forced more car companies to choose to equip passenger cars with automatic driving functions.

Starting in 2021, as one of the core elements of this competition, lidar has once again become the selling point of car companies. According to CICC statistics, at least 20 of the new models launched in 2021-2023 have been announced to be equipped with lidar.

The Chinese market has a great enthusiasm for lidar, so that its financing and fixed-point models have mushroomed.

The lidar is constantly rolling inward

As the first of the three core links of intelligent driving, "perception, decision-making, and execution", lidar is known as the "eye of machinery and equipment" and is one of the key technologies for high-level automatic driving.

In fact, at the end of 2020, some mass-produced passenger cars will begin to be equipped with lidar. In 2021, the pace of lidar "getting on the car" is amazing. Compared with the more mature millimeter-wave radar, lidar is difficult to land, and the only mass production model equipped with lidar in the world before 2020 is the Audi A8 released in 2017.

Previously, Weilai co-founder Qin Lihong had revealed that "compared with the lidar carried by Weilai, other families are weakly exploded", and at the NIO day in January last year, the ET7 in its product matrix for automatic driving exploration revealed this puzzle: ET7 is equipped with image-level ultra-long-range lidar developed by Innovusion and NIO, with a 120-degree ultra-wide viewing angle and ultra-high resolution, which can detect up to 500 meters.

Weilai's in-house Innovusion has set off a new wave of fixed-point lidar in the industry, and since then, the entire industry chain has begun to rub for high-level automatic driving.

Last year, Xiaopeng, BAIC Jihu, SAIC Zhiji, SAIC Feifan, GAC Aian, Changan Avita, Great Wall Salon, Ideal, Gaohe, Weima and other new car-making forces and traditional car companies have announced that new models that will be mass-produced or delivered will be equipped with lidar; on the other hand, international traditional giants such as Honda and Nissan have also announced that they will be equipped with lidar on new mass production vehicles.

"In the autopilot industry, except for Tesla, the vast majority of car companies tend to use multiple redundant perception schemes such as lidar + millimeter wave radar + cameras." Mr. Gao of a head car company said.

"For this part of the members of the fusion perception school, lidar and millimeter-wave radar have evolved from 'loading or not loading' to 'loading how many'."

According to incomplete statistics, in addition to the Weilai ET7, Gaohe HiPhi Z, SAIC Zhiji L7, etc. equipped with a lidar, Xiaopeng G9, SAIC Zhiji LS7 and other models are equipped with two lidars, gac Aion LX Plus, Avita 11, WM M7 and other models equipped with three lidars.

At the Guangzhou Auto Show in November 2021, Great Wall's Sharon Motors launched its first mass-produced model, the Mech Dragon, and declared it the world's first model equipped with four lidars. Sharon Cars shouted domineeringly before the exhibition: "Don't talk under 4."

The lidar joined hands with the automakers and rolled up.

Visualism and Fusion Perception

Since the development of autonomous driving technology, two gate schools have been formed, one is the visual school represented by Tesla, and the other is the fusion perception school represented by other car companies.

In May 2021, the radical Tesla of self-driving gradually replaced the Model 3 and Model Y hardware from the factory with a pure vision system, and said that its electric vehicles would use pure vision systems to work and completely abandon the on-board radar.

Currently, the Model 3 and Model Y's driver assistance features on sale only rely on the "TeslaVisiom" vision system consisting of eight 1.2-megapixel cameras.

At the T-talk offline exchange meeting, Tesla said that sometimes the information collected by millimeter-wave radar may be a kind of interference for the automatic driving system, rather than forcibly integrating the reflected wave information collected by millimeter-wave radar with the visual information collected by the camera, it is better to make the pure visual solution to the extreme.

Musk said that if the information from the radar and camera contradicts each other, the system will be more difficult to choose. Instead of letting the two pull each other's hind legs, it is better to choose only one and make it to the extreme. The domestic manufacturers' lidar was criticized by him as "only for fools".

"Tesla's technical capabilities are indeed the industry benchmark at present, and the reason why it dares to remove millimeter-wave radar and not use lidar as a source of information is because Tesla image processing can achieve depth information and high reliability." Mr. Gao said, "Nevertheless, pure vision is not a replicable scheme. At present, the technical capabilities of other companies are not enough to support, so in the face of actual demand, they can only be forced to choose to use a variety of sensors to supplement each other's information. The depth information that the camera cannot capture can only be obtained with lidar. ”

Mr. Gao told the author that compared with a single camera, radar has more dimensions when collecting data, so the information transmitted by radar is richer than that of the camera, and lidar has its own light source, so its performance is significantly better than that of cameras in the case of insufficient light or poor weather conditions.

In addition, Mr. Gao expressed this view: the use of pure vision, Tesla or for cost considerations.

"The cost of fusion perception is too high, a millimeter wave is hundreds of yuan, lidar is thousands of yuan, compared to the camera is cheap, so Tesla still occupies a price advantage on the whole vehicle."

At present, the price of lidar ranges from a few thousand dollars to tens of thousands of dollars, but the cost shows a downward trend.

"In the future, the price of lidar will decline after mass production, and Tesla may not use lidar."

When talking about the development trend of the two, Mr. Gao said that under the premise of the same cost or little difference, the two directions may present the possibility of coexistence: "The coexistence of two technologies is also possible, and industrial integration is also possible." ”

The distance of autonomous driving to develop to L4 or even L5, there is still a long way to go, whether it is the north slope to the top, or the south slope to the top, risk and difficulty coexist, the key is to see who can persist to the end.

The night before the mass production of the deer

At present, more and more players are entering the lidar track.

In addition to the domestic Hesai Technology, Sagitar Juchuang and other specialized manufacturers, Huawei, DJI and other technology companies have also built radar; while foreign countries have Velodyne, Luminar, Innoviz and other manufacturers to force, Amazon also announced the acquisition of 16% of the shares of LiDAR company Velodyne, adding to the automatic driving business.

Since 2021, a number of car companies have joined hands with lidar suppliers and successively announced the timetable for mass production.

Innovusion co-founder and CEO Bao Junwei also said on NIO Day2021: "Next year (2022) we are sure that we can get on the car, and the probability of other friends getting on the car is also relatively large, so that the first year of lidar mass production will really arrive." ”

Recently, there has been a rumor inside and outside the industry that "lidar is about to usher in an explosive period".

"I think it's okay to call 2022 the first year of mass production," Mr. Gao replied, "This year there will be radar loading, which will serve as a sign of the beginning of the development of the industry, but there is still a long way to go before large-scale mass production." ”

As far as the entire industry is concerned, there is no doubt that lidar has broad prospects, so large-scale mass production has become an unavoidable topic.

"The key node of large-scale mass production is still closely related to the development speed and rhythm of autonomous driving technology, after all, demand is the most basic business rule", Mr. Gao said, "2025 will be the time node for the large-scale application of L3-level autonomous vehicle mass production, and lidar should also follow if it wants to achieve large-scale mass production." I think the price of lidar will remain high until 2025. ”

The industry believes that the current lidar allocation rate is still not high, and the downstream demand is relatively low. Even if the pace of the demand side keeps up, the supply side production rate trapped by technology is relatively slow.

To truly realize the scale of lidar on the train, in the final analysis, it is necessary to break through technical barriers and the developability of technology.

Add group tips

LEGAL COUNSEL: Yingke Law Firm

Intensive Reading