Reprinted from Economic Observer

Economic Observer Network reporter Zhou Ju said, "Our human eyes are much more powerful than our current camera technology. Based on our current camera technology, it is a bit unfair to complete assisted driving or autonomous driving, just like how we compete with a sixth grade elementary school student against a college student." Bai Jian, Vice President of NIO Automotive Hardware, stated in an online salon held on June 22 that in the foreseeable future, NIO will not choose a purely visual perception route, but will adhere to a multi-sensor fusion route, with an important sensor being the LiDAR.

NIO's attitude towards LiDAR and sensor fusion is a common attitude among many domestic car companies. At present, there has been controversy over the perception scheme for autonomous driving in intelligent vehicles, which can be roughly divided into two schools: the pure visual school, represented by Tesla; Another type is the multi-sensor fusion genre that includes LiDAR. Mainstream smart car manufacturers in China, such as NIO, Ideal, and Xiaopeng, all belong to the latter.

At this online salon, NIO introduced for the first time the LiDAR equipped on its flagship pure electric sedan ET7, as well as the logic behind its assisted driving perception scheme and route selection. On March 28th this year, the NIO ET7 announced the start of delivery, and the LiDAR was first mass-produced and equipped on this car.

Lidar solutions are gradually becoming mainstream

Bai Jian stated that human eyes are more precise than cameras, and the eyes and brain can instantly obtain depth information. However, in a purely visual approach, cameras can only provide partial depth information, and the accuracy decreases as the distance increases. In addition, the focal length of the human eye can be changed at any time, allowing for quick switching between panoramic and instant focus. However, current camera technology is unable to achieve the powerful functions of the human eye, so it is necessary to use more powerful LiDAR for fusion perception.

Tesla's approach and perspective are completely different, as it has always adopted a purely visual approach. In Tesla CEO Musk's view, radar sensors are not only useless, but also an obstacle to autonomous driving. In an interview with the media, he said frankly, "The safety probability of the pure vision scheme will be higher than that of the vision+radar, and the vision scheme is already quite good. If the radar is used again, it will provide the (automatic driving) system with excessive and unnecessary information, and may also interfere with the signal."

There is still ongoing debate in the industry about who will have the upper hand over the two routes. Although LiDAR has better perception effects, its cost is relatively high. "Lidar is indeed very expensive, but it is not as expensive as our foreign friend's boss (Musk) mentioned. China's supply chain capabilities and design level are particularly good at doing (cost reduction) this thing. With the scale of mass production, its sinking space and strength are still very considerable." Bai Jian said.

With more car companies investing in LiDAR, Tesla's pure visual route is also showing signs of wavering. Recently, it was reported that Tesla has officially submitted relevant materials for millimeter wave radar to the Federal Communications Commission (FCC), which means Tesla may be designing millimeter wave radar. According to the documents provided by Tesla to the authorities, this high-resolution radar will be used for imaging, combined with cameras for autonomous driving detection solutions.

Musk once mentioned in an interview that if Tesla had "ultra-high resolution radar," it was not ruled out that they would use it. "A visual solution using high-resolution radar would be better than a pure visual solution."

NIO follows a self-developed path

Among many car companies that support LiDAR, NIO also appears to be unique, with most companies directly using third-party suppliers, while NIO adheres to its self-developed route. It is reported that the LiDAR equipped on the ET7 was jointly developed by NIO and Tudatong. Bai Jian stated that this mode has brought several obvious benefits to NIO.

Firstly, it shortens the development time. "In the past, sensors in the automotive industry were mostly developed by Tier 1 first, and then handed over to the host factory for installation, testing, and acceptance. If there were no issues with acceptance, mass production would begin. However, from a time perspective, because Tier 1 requires one or two or even two years of self-development, and then handed over to the host factory for installation, installation, and certification, it may also take several years, and the entire cycle is very long," said Bai Jian.

Through joint development, NIO stated that the development cycle of LiDAR has been greatly shortened, with a LiDAR taking less than two years from project approval to launch.

In addition, joint development facilitates deep customization and modification of products by host manufacturers. "In the previous mode, Tier 1 was developed, and it was difficult for OEMs to make some deep customization or modifications. You neither had the time nor the ability to do this. Therefore, it was difficult to maximize the performance of ECU or sensors." Bai Jian stated that in the case of joint development by both parties, NIO could tailor products such as ET7 to achieve optimal wind resistance coefficient and various performance indicators of the new car.

NIO also stated that the joint development of LiDAR can improve its supply chain, reliability, and cost.

As a new car manufacturing enterprise, NIO has emphasized full stack self-development since its inception, but Bai Jian also pointed out that NIO cannot do everything. "We are first and foremost a host factory or a car factory, and we cannot do everything in the industry. We believe that there is a very large ecosystem in the industry, and only openness is the solution. Moreover, from the perspective of talent and resources, it is impossible for one company to fully cover it," said Bai Jian.

According to the data, the recently released Ideal L9 by Ideal Automobile uses the AT128 LiDAR from Hesai Technology, while the Xiaopeng G9 from Xiaopeng Automobile uses the Sagitar Juchuang product. The first concept car of Jidu Automobile, ROBO-01, is also equipped with two LiDARs from Hesai Technology.

In the automotive industry, LiDAR has become one of the competition focuses for new car companies, and at the same time, an arms race around LiDAR has also begun. According to incomplete statistics, the number of LiDARs for each new car is 1 for NIO, 2 for Xiaopeng, 3 for Weima, and 3 for Jihu, while the number for Salon Automotive Mechatron is 4. But is it better to have more LiDARs?

Bai Jian believes that "when dealing with complex scenarios such as forward, high-speed, and processing, LiDAR is needed (or in other words, LiDAR is particularly needed), so we chose this forward LiDAR. There are some lateral LiDARs, which are mostly used in unprotected left turns and Cut-ins, where the vehicle speed is generally low. In this case, the vision system of (NIO) Aquila can provide sufficient depth information."

This article is for academic sharing only. If there is any infringement, please contact us to delete the article.