Author / Yuan Yanjie
Recently, TEL AVIV, Israel – LiDAR system supplier Liangdao Intelligent and LiDAR listed company Innoviz (Nasdaq: INVZ) held an online media conference and jointly announced that the two sides have signed a strategic cooperation agreement and formally established a strategic partnership.
（InnovizTwo）
Under the agreement, the two parties will explore the possibility of customizing Innoviz LiDAR products for automakers in China, developing perception solutions, and conducting system verification and data collection to accelerate the roll-out of a serialized production solution for High Resolution LiDAR based on InnovizTwo.
Nowadays, at the stage when the route of autonomous driving technology is still inconclusive, at this conference, Liangdao Intelligent CEO Ju Xueming and Su Shuping, general manager of Innoviz China, also expressed their views on everyone's questions about lidar and related topics.
Mass production is still not possible this year
On the question of whether lidar can achieve large-scale mass production and on-board, Su Shuping believes that the first step is to define a "large-scale" magnitude - "at least 500,000 vehicles per year."
According to this definition, she admits that lidar still cannot achieve mass production this year: "In 2022, we are still in the stage from 0 to 1, and whether we can cross the threshold of 1 is still debatable; and the real wait for large-scale mass production should be after 2024 to 2025."
Indeed, from the timeline point of view, as early as 2018, Valeo released the world's first car-grade lidar - Valeo SCALA, and installed on the Audi A8. But obviously, this car in the practical application level, did not achieve the desired effect.
Because in the view of The Drama Xueming, the completion of the test is only the first step in the long march: "Can the first generation of mass-produced lidar in the year be able to open the L3 function without speed limitation in HWP (High Way Pilot)?" This is actually challenging. To get lidar to open up more scenes and more functions, it takes a long time to accumulate."
Regarding the time point of large-scale mass production, Ju Xueming and Su Shuping hold the same view: it may start from 2024 to really increase the volume, and there will be a relatively large increase in 2025.
Difficulties are not just "three mountains"
The "three mountains" in front of lidar are: cost, volume and vehicle specification.
One of the challenges encountered in the domestic market is the price, which CES just launched this year. Su Shuping said frankly: "If you want to balance high resolution, large FOV and high FPS, you can't achieve the lowest price. In addition, compared to the small size of millimeter-wave radar and cameras, the size of lidar is far from enough."
(InnovizOne B sample point cloud)
Getting in the car is easy, it's hard to use.
With more and more official announcements of mass production projects, the future of lidar is undoubtedly bright, but its way forward is also very tortuous: "When we really sink our hearts into this, the difficulty is not only three mountains."
Su Shuping said that the market's requirements for lidar are much higher than those of cameras. In terms of distance, more than 200 meters is the basic requirement of the lidar perception distance; for FOV, customers generally require lidar to do 120 °, or even 150 °; at the same time, in terms of resolution, the earliest lidar is generally measured by the number of lines, like the resolution of 16 lines and 32 lines for mass production vehicles. "Now it's 128 lines, and some can even reach 300 lines."
In addition, in order to make the product really land, in addition to technical problems, there are also production problems. How to quickly achieve large-scale, high yield, and good consistency of lidar is also a big challenge. In this regard, The drama Xueming said bluntly: "Whether it is the earliest Audi project or the current BMW project, none of them is simple."
Self-research and cooperation are the way out
"Only through self-research and cooperation can we build a balance that satisfies the market's pursuit of high performance while reducing the price to the right state." Su Shuping said that Innoviz's lidar is based on MEMS technology, in which the stability, reliability and high performance of MEMS galvanometers are contradictory. The self-developed MEMS galvanometer, receiver and signal processing chip are the key to a good balance. For example, LiDAR's B prototype, The InnovizTwo, has 30 times more performance and a 70% lower cost than the first generation.
(InnovizTwo related data)
In view of how to reduce costs, Su Shuping believes that technological innovation is the core of reducing costs.
Using the example of InnovizTwo, she said that compared to The InnovizOne's four-way optical path, InnovizTwo has only one optical path, thanks to the technological breakthrough brought about by the self-developed core components.
Ju Xueming said that chipping is a way to ensure future cost reduction. As the amount of 905 nanometers and silicon-based use rises, there is room for further price declines.
In order to further promote the mass production application of lidar products in the Chinese market, Liangdao Intelligent and Innoviz will carry out in-depth cooperation in the Chinese market based on their respective technologies and resources. Last month, after more than three years of operations in China, Innoviz has officially set up a subsidiary in China, Yingnovos (Shanghai) Technology Trading Co., Ltd., and the new company will continue to be led by Su Shuping.
In fact, Liangdao Intelligence's cooperation with Innoviz has a long history. In 2018, BMW selected Innoviz's InnovizOne solid-state LiDAR sensor, which was procured and manufactured by Magna International to develop its LiDAR for serial production. Liangdao Intelligence is responsible for collecting domestic data in this cooperation.
Last year, Liangdao Intelligent and Shanghai Auto Inspection jointly established a lidar point cloud joint test laboratory, which provides lidar point cloud quality evaluation for domestic and foreign host manufacturers and lidar manufacturers, based on the evaluation method of China's automotive industry group standards, can carry out more than 30 technical evaluations of the basic performance of vehicle lidar point cloud, and output professional third-party evaluation reports for customers.
Innoviz was the first lidar tested in a joint laboratory. After boarding the car, lidar will be further tested according to the needs of customers, in the process, Liangdao Intelligence will exert its own data collection and data verification capabilities. Drama Xueming also revealed at the press conference: "In the cooperation with Innoviz, some technologies are open to each other."
(InnovizOne test data)
In the future, OEMs have no reason not to use lidar
At present, the dispute over the lidar technology route has always been a hot topic in the industry, and Toyota's addition of pure visual autonomous driving technology routes has recently attracted the attention of the industry.
In the future of intelligent travel, the controversy will continue whether it will continue to bloom or whether some kind of technical route will eventually win. In this regard, the view given by Ju Xueming is: "Stacking or magic, at present, it is difficult to distinguish between victory and defeat for the time being." At least for a long time, there will still be multiple technical routes, and the relationship between them is a combination rather than a substitution."
Chen Na, CSO of Liangdao Intelligence, clearly put forward the recommended combination method: "In the future, the way we recommend is to equip two to four auxiliary lidars on the basis of a high-precision main sensor, and Flash lidar is used as an angle radar." This may be the mainstream direction in the future."
(Liangdao intelligent boarding combination scheme)
Su Shuping said: "What the future will look like, there is still a lot of uncertainty in the market. Lidar not only has to consider scanning, other whether it is a light source or a Detector, each core component will have a different technical route, and the type of combination is very complex. In short, so far, lidar is an indispensable sensor."
For the pure visual road, Su Shuping said that at present, some companies may take the route of pure vision for two reasons, one is that it is more sensitive to price, and the second is because lidar is not mature enough, customers have concerns about installing lidar at a high price, and once the threshold of L3 is crossed in the future, pure vision is not enough, because it cannot replace the main sensor of L2 - the driver.
Drama Xueming predicted: "The future of lidar sensors will become blurred to some extent by the definition of cameras. It is possible that lidar can perform both millimeter wave and lidar, or that lidar itself can output images. In this case, it is a new type of sensor at a low cost. From the perspective of the depot architecture, safety redundancy must be considered, so lidar is indispensable, and OEMs have no reason not to use lidar."
At present, the lidar industry is fiercely competitive, and in the end, Ju Xueming gave his own views: "After the real large-scale mass production of lidar in the future, it is impossible for the world to accommodate hundreds of lidar startups, but there will be a head gathering." As just mentioned, because the actual application scenarios are different, multiple technical routes cannot be replaced by each other, so there should be more than ten players in the end, and it depends on who can achieve the best chip design, the fastest iteration, the lowest cost, and the best performance."
About Auto Byte
Auto Byte launches the Automotive Technology Vertical Media for The Heart of Machines, focusing on cutting-edge research and technology applications in the fields of autonomous driving, new energy, chips, software, automobile manufacturing and intelligent transportation, and helping professional practitioners and relevant users in the automotive field understand technology development and industry trends through technology to gain insight into products, companies and industries.
Welcome to follow the stars and click Likes and Likes and Are Watching in the bottom right corner.