Automotive Sensor - LiDAR Chip Industry Chain
[Copy link]
The upstream of the LiDAR industry chain includes lasers, light sources, optical components, galvanometers, analog chips, FPGA chips, etc.; the midstream of the industry chain includes LiDAR hardware manufacturers; the downstream industry chain is mainly divided into autonomous driving, advanced driver assistance, service robots and Internet of Vehicles industries according to application areas.
Overall, the entire laser radar industry chain has shown outstanding characteristics of rapid development, high technological level, strong innovation ability, and broad market prospects. From the perspective of comparing foreign and domestic industry chains, foreign upstream laser radar companies have started earlier and have deeper accumulation, especially in the field of underlying optoelectronic devices and chips. Foreign and domestic downstream laser radar companies are comparable in product performance and commercialization progress. Domestic upstream suppliers have also developed rapidly in recent years and are expected to gradually catch up with overseas upstream companies.
(1) Upstream suppliers
The upstream industrial chain of the lidar industry mainly includes laser emitters, photodetectors, FPGA chips, analog chip suppliers, and optical component manufacturers and processors.
Laser emitters and photodetectors are important components of LiDAR. The performance, cost, and reliability of laser emitters and detectors are closely related to the performance, cost, and reliability of LiDAR products. Moreover, the system design of LiDAR will put forward customized requirements for the specifications of laser emitters and detectors. In-depth cooperation with upstream suppliers to customize laser emitters and detectors will help improve the competitiveness of products.
According to the different gain media, lasers can be divided into five categories: gas lasers, solid-state lasers, fiber lasers, semiconductor lasers (laser diodes) and liquid lasers. EEL and VCSEL are both semiconductor lasers, and fiber lasers mainly use semiconductor lasers as pump sources. These three are the more mainstream laser products at present.
Comparison of VCSEL and EEL Process
EEL has the performance advantage of high power density, but its light-emitting surface is located on the side of the semiconductor wafer. During use, it needs to go through the process steps of cutting, flipping, coating, and re-cutting. It is extremely dependent on the manual adjustment technology of the production line workers, with high production costs and difficult to ensure consistency. In addition, EEL can only fully generate lasers after the wafer is cut, and it cannot be tested during the production process. The light-emitting surface of VCSEL is parallel to the semiconductor wafer, and has the characteristics of surface light emission. The emission beam is narrow and round. The laser array formed is easy to bond with planarized circuit chips. The precision is guaranteed by semiconductor processing equipment, and it is easy to integrate with silicon material micro lenses of surface processes to improve beam quality. In recent years, many VCSEL laser companies at home and abroad have developed multi-layer junction VCSEL lasers, which have increased their light-emitting power density by 5-10 times, which has made it possible to develop long-range lidar using VCSEL.
The major laser transmitter manufacturers in the industry include OSRAM, AMS, Lumentum and others from abroad; Shenzhen Ripple Optoelectronics Co., Ltd., Changzhou Zonghuixing Optoelectronics Technology Co., Ltd. and others from China.
Foreign suppliers have been working in the laser emitter and detector industry for a long time, have more practical experience and advantages in product maturity and reliability, and have a wider customer base. Domestic suppliers have developed rapidly in recent years, and their product performance has basically approached the level of foreign supply chains. There are already domestic laser emitters and detectors that have passed automotive certification (AEC-Q102). The automotive standardization of components is the basis for the realization of automotive-grade lidar, and domestic suppliers can meet this demand. Compared with foreign suppliers, domestic suppliers have greater flexibility in product customization and certain price advantages.
②Photodetector
PD (photodiode) is a common choice for lidar detectors due to its low cost. Its disadvantage is that it has low sensitivity and is only suitable for short-distance detection. For long-distance detection, a high-gain APD (avalanche diode) is required. APD is divided into linear working mode and Geiger working mode. The Geiger working mode has the highest gain and only outputs a 1 or 0 level signal. It has high sensitivity and is called SPAD (single photon avalanche diode). SPAD can operate independently at a single point or form an array (for table distinction, this report uses SPPC to refer to SPAD array). Each SPAD in the SPPC works independently.
SiPM (Silicon Photomultiplier, also known as MPPC) is another form of SPAD array, which is composed of multiple independent SPAD sensors in parallel, and the output signal will have different amplitude levels. According to ON Semiconductor data, the usage rates of PD and APD in the current LiDAR market have declined to varying degrees, while the usage rates of SPAD, SPPC and SiPM are continuously increasing. Although SPAD has higher sensitivity and conversion efficiency, it is subject to high cost and low mass production consistency, so APD is still the mainstream at present.
③Lidar chip
I.FPGA chip
FPGA chips are usually used as the main control chips of LiDAR. The mainstream foreign suppliers include Xilinx, Intel, etc. The main domestic suppliers include Unigroup Guoxin Co., Ltd., Xi'an Zhiduojing Microelectronics Co., Ltd., etc. The product performance of foreign suppliers is far ahead of that of domestic suppliers, but the logic resource scale and high-speed interface performance of domestic products can also meet the needs of LiDAR.
However, FPGA is not the only choice for the main control chip of LiDAR. High-performance microcontrollers (MCU) and digital signal processing units (DSP) can also be used instead. The international mainstream suppliers of MCU include Renesas Electronics and Infineon, and the mainstream suppliers of DSP include TI (Texas Instruments) and ADI (Analog Devices).
II. Analog Chip
Analog chips are used to build key subsystems such as light control, photoelectric signal conversion, and real-time processing of electrical signals in LiDAR systems. The main international suppliers of analog chips include TI (Texas Instruments), ADI (Analog Devices), etc. Domestic analog chip suppliers mainly include Silergy Semiconductor Technology Co., Ltd., Shengbang Microelectronics (Beijing) Co., Ltd., etc. Foreign suppliers have accumulated a long history in this field, with advanced technology, sufficient production capacity, and high maturity, and are leaders in the industry. Domestic suppliers started later than foreign suppliers, and there is generally a certain gap from product richness to technical level, especially for automotive products.
Semiconductor industry chain
In terms of optical components, LiDAR companies generally conduct independent R&D and design, and then select processing companies in the industry to complete the production and processing procedures. The technical level of the domestic optical component supply chain has completely reached or exceeded the level of foreign supply chains, and has obvious cost advantages, which can completely replace foreign supply chains and meet product processing needs.
Progress in Chip Commercial Applications
(1) Automotive field
Based on the concepts of safety first and mass production, the OEMs have a relatively conservative technical route, and most of them gradually realize intelligent driving through the expansion and improvement of ADAS functions. ADAS mainly refers to the advanced assisted driving functions such as adaptive cruise control and automatic following by passenger cars equipped with laser radar. Compared with the laser radar for intelligent driving, the laser radar used in ADAS has higher requirements for vehicle-standard mass production capacity and reliability, and is more sensitive to cost.
The business model of technology companies (Waymo, Momenta, etc.) is mainly based on selling intelligent driving solutions. Because technology innovation companies have a complete talent pool, strong algorithm strength, and relatively radical technical routes, they directly start to develop L4/L5 intelligent driving technology. LiDAR is the core sensor for realizing L4/L5 intelligent driving technology. Many intelligent driving technology companies at home and abroad use the point cloud data output by LiDAR as the main basis for decision-making. As the main sensor, the 360° mechanical rotating LiDAR is generally placed on the roof of Robotaxi or Robotruck, and a vehicle is usually equipped with multiple LiDARs, covering both sides of the vehicle body or taking into account long and short distances to meet the detection needs of complex road scenes. Taking Inceptio Technology as an example, the L3-level Robotruck jointly developed with Dongfeng Commercial Vehicle is equipped with 3 LiDARs, of which the main radar uses a RoboSense MEMS LiDAR, and the two corner radars use the MEMS products of Yijing Technology.
① OEM mass-produced models
At present, both OEMs and LiDAR manufacturers are working hard to promote the mass production of LiDAR. Since 2021, the mass production of LiDAR has accelerated. During the Shanghai Auto Show and the Guangzhou Auto Show, many models announced that they would be equipped with LiDAR, and the criticality and necessity of LiDAR were further confirmed. The mass-produced LiDAR models Xiaopeng P5, BAIC Alpha S, and WEY Mocha were launched one after another, and domestic OEM manufacturers took the lead in installing LiDAR on vehicles. On April 14, 2021, the Xiaopeng P5 model equipped with DJI Livox LiDAR was released; on April 18, 2021, the BAIC Alpha S equipped with Huawei LiDAR was released. Most car companies that are expected to be equipped with LiDAR are expected to mass-produce in 2021-2022.
LiDAR-related automotive standards
Source: Minsheng Securities
At present, the overall design direction of automotive-grade LiDAR products is low cost, high performance, high integration, and solid state. Each manufacturer has selected different technical solutions based on its own technical reserves in order to meet the needs of OEM manufacturers. LiDAR manufacturers form special solutions by combining the design of LiDAR transmitting system, receiving system, information processing system, and scanning system.
As can be seen from the table below, domestic and foreign OEMs and new car manufacturers all have plans to install LiDAR in 2022, including new models and modified models. The number of LiDARs installed on a single vehicle ranges from 1 to 5, with 1 (main radar) and 3 (1 main radar + 2 corner radars) being the mainstream solutions. There are as many as 11 hardware partners, which shows that the leading radar manufacturers have no obvious competitiveness and the market is still in a relatively early stage of open competition for market share.
Statistics of models equipped with LiDAR
Source: EoAuto, official websites of various OEMs
In addition to the plan to install LiDAR in cars, domestic and foreign OEMs have also invested in LiDAR manufacturers in recent years. According to incomplete statistics, the investment events of OEMs are summarized as follows:
OEM investment in LiDAR cases
Source: Compiled by Dezaihou Capital
②Leapfrog camp
The leapfrog camp is mainly composed of autonomous driving technology startups. Due to the different scenarios covered in the early stages of development, it is different from the gradual development route of traditional OEMs and new forces from ADAS to fully autonomous driving. This camp generally adopts a "one-step" strategy and directly develops L4 intelligent driving technology with lidar as the core sensor. The application scenarios are relatively conservative, mainly in closed or semi-closed scenarios such as factories, ports, mining areas, airports, and highways.
Unlike passenger car OEMs, which have strict requirements on whether the LiDAR they choose has passed the automotive grade certification, autonomous driving solution providers pay more attention to product performance and cost performance. In addition, considering the response speed and customer customized development needs, domestic autonomous driving companies prefer to use products from local LiDAR hardware suppliers.
LiDAR suppliers of mainstream autonomous driving companies (partial)
Source: Public information
(2) Smart Transportation
In addition to the field of intelligent driving, the application areas of lidar are also constantly expanding, including pre-installed advanced assisted driving represented by automobile OEMs and Tier 1, obstacle avoidance and navigation systems represented by intelligent service robots, and intelligent transportation vehicle-road collaborative applications that have emerged with the gradual popularization of 5G technology, all of which have brought a broader market for lidar.
Roadside LiDAR has two main functions: one is to collect high-precision maps, and the other is to monitor road traffic in real time. The roadside traffic field is mainly dominated by relevant government departments. The requirements for automotive-grade integration are relatively low, but the requirements for algorithms are very high. As a roadside sensor, LiDAR needs to monitor and perceive road users. Currently, mechanical rotating LiDAR is the main application product. The application scenarios are mainly highways and intersections.
①Smart highway market
my country's current existing expressway mileage is 161,000 kilometers. Assuming that the annual mileage of newly built expressways remains unchanged at 7,500 kilometers, and based on the calculation that a single lidar device covers 200 meters, four lidars are required per kilometer in the expressway scene. It is estimated that the penetration rate will reach 10% in 2025, and the market size will be approximately 3.7 billion yuan in the next five years.
②City intersections
There are currently about 300,000 urban intersections in my country. Assuming they are arranged diagonally, two lidars will be required at each intersection. If the market penetration rate reaches 10% in 2025, the total market space in the next five years will be approximately 2.3 billion yuan.
| 
提升卡
变色卡
千斤顶
京公网安备 11010802033920号
