Foreword: [Core Watch] is an in-depth series of columns produced by the editorial department of electronic enthusiasts. The purpose is to use the most intuitive way to make readers understand the structure of the electronic industry as soon as possible, to clarify the various links of the upper, middle and lower reaches, and to quickly understand the major details. Industry status in sub-segments. In this issue of [Core Observation], we will sort out the hot lidar industry in recent years, and analyze the lidar industry chain from upstream components, mid-stream and downstream terminal manufacturers and other parts.

One: Lidar: the core sensor of autonomous driving

Lidar is a radar system that detects the location, speed, structure and other characteristics of a target by emitting a laser beam. Similar to the principle of other radar systems, lidar transmits a detection signal (laser beam) to the target, and then compares the signal reflected by the detection target with the original signal at the time of transmission, and obtains relevant information of the target through a certain algorithm, including the target distance. , bearing, speed, and even shape. The advantage in perception of the external environment also makes lidar favored at the beginning of the development of autonomous driving technology.

Looking back at the development history of lidar, it was first used in the fields of environmental surveying and mapping, but since 2005, a founder of a speaker company who is “not doing a good job”, a lidar with an innovative structure, an autonomous driving competition, let the laser Radar has truly entered a new chapter.

The main business was originally Velodyne, a speaker. Its founder David Hall participated in the DARPA Autopilot Challenge in 2004. At that time, David Hall modified a pickup truck and used a panoramic camera as a sensor for autonomous driving. Interestingly, in the first DARPA Autonomous Driving Challenge in 2004, none of the participating vehicles finally completed the competition.

But in this competition, David Hall learned about the existence of lidar for the first time in the communication with other contestants. In the following year, David Hall innovatively developed a 360° rotating lidar, which was applied to self-driving cars again in 2005. Although he eventually retired due to a mechanical failure of the vehicle, the huge sensor installed on the roof attracted widespread attention for the lidar for a while, and Velodyne has since transformed to concentrate on the development of lidar.

In the 2007 DARPA Urban Challenge, 6 of the 7 cars that finished the race were equipped with lidar. Since then, lidar has officially established an inseparable relationship with autonomous driving.

In 2018, the Audi A8 offered the option of lidar and became the world’s first mass-produced passenger vehicle equipped with lidar. However, since the regulations on autonomous driving in countries around the world had not yet been perfected, this model was used in most regions. Level 3 automatic driving assistance cannot be enabled.

In 2021, which is considered to be the first year of LiDAR, Xiaopeng P5 became the first pure electric smart car equipped with LiDAR, and then a number of mass-produced models equipped with LiDAR were launched, including LucidAir, Weilai ET7, etc. . With the continuous penetration of ADAS, robotaxi, etc., the demand for lidar will also usher in a stage of rapid growth.

IDC predicts that the total global shipments of autonomous vehicles will increase from 27.735 million in 2020 to 54.247 million in 2024, the penetration rate is expected to exceed 50%, and the compound annual growth rate from 2020 to 2024 will reach 18.3%, of which L3 level shipments in 2024 may reach about 690,000 units.

In the first quarter of 2022, the penetration rate of L2-level autonomous passenger vehicles in China’s autonomous vehicle market was as high as 23.2%, a huge increase from 7.5% in the first quarter of 2021.

According to Sullivan’s forecast, driven by factors such as the expansion of the unmanned fleet and the increase in the penetration rate of lidar in ADAS, the overall lidar market is expected to show a rapid development trend. By 2025, the global market size is 13.54 billion US dollars (about 91.4 billion yuan); among them, the scale of China’s lidar market will reach 4.31 billion US dollars (about 29.1 billion yuan).

Two: Lidar technology route, core components summary

  1. Technical route: scanning method, ranging method

First of all, in terms of scanning methods, lidars are currently mainly divided into: mechanical rotary, prism, rotating mirror, MEMS galvanometer, OPA, Flash, etc. Among them, OPA and FLASH are the technical routes to realize solid-state lidar, while the current mainstream hybrid solid-state lidar uses several technologies such as prism, rotating mirror, and MEMS galvanometer.

Mechanical Rotation: Omnidirectional coverage is obtained by rotating the entire laser transmitter module and receiver module 360° laterally. The laser beams are arranged vertically to form a surface, and the number of lines such as 16 lines and 64 lines is the number of vertically arranged laser beams. The more the number, the higher the resolution and the greater the amount of information. Therefore, we are testing the autonomous driving road. This type of lidar is often seen in vehicles of . However, the mechanical rotary lidar is large in size, complicated in debugging, high in cost, difficult in mass production, and difficult to meet the requirements of vehicle regulations due to the characteristics of mechanical structure, and has a short lifespan.

Major manufacturers: Velodyne, Hesai Technology, Sagitar Juchuang

Rotating mirror: Rotating mirror is divided into one-dimensional rotating mirror and two-dimensional rotating mirror. The one-dimensional rotating mirror reflects the laser light in different directions through a rotating polyhedral mirror, thereby achieving a certain field of view; the two-dimensional rotating mirror, as the name implies, integrates two rotating mirrors, one polygonal prism is responsible for lateral rotation, and one It is responsible for vertical flipping, which can realize scanning in two directions and dimensions with one laser beam. Rotating mirror lidar is small in size, low in cost, and has the same effect as mechanical lidar. It is currently the mainstream route of hybrid solid-state radar. But after all, there is also a mechanical structure, and the frequency of mechanical movement is high, and the lifespan is also not ideal.

Major manufacturers: Valeo, Huawei, Velodyne, Leishen Intelligence

MEMS galvanometer: A mirror integrated on a silicon-based chip is used to vibrate with a certain harmonic between a pair of torsion bars in the front, rear, left, and right to reflect the laser beam to different angles to achieve scanning within a certain range. This technology gets rid of the assembly of mechanical moving parts, improves reliability, and makes it easier to achieve mass production. At the same time, the size of the lidar system can be reduced, the number of laser transmitters and detectors can be reduced, and the cost can be greatly reduced. However, the limited optical aperture and scanning angle limit the ranging and field of view. Large field of view requires splicing, which requires higher algorithms. In addition, there are some problems with impact resistance and reliability.

Major manufacturers: Sagitar Juchuang, Innoviz, Valeo

Rotating mirror + MEMS galvanometer: adding a galvanometer on the basis of a two-dimensional rotating mirror, the rotating mirror is responsible for the horizontal direction, and the galvanometer is responsible for the vertical direction, which can achieve a larger scanning area and a higher frequency, but the price is also higher.

Major manufacturers: Innovusion, Luminar

Prism: By controlling the relative rotation speed of the two wedge prisms, the laser beam is refracted twice to achieve regional laser scanning coverage. The scanning pattern accumulated by the prism lidar is shaped like a petal. The number of scans at the center point is dense, and the edge of the circle is relatively sparse. The scanning time is long to enrich the image, so it is necessary to add multiple lidars to work together to achieve a higher effect. Prisms can achieve high-precision and long-distance detection by increasing the laser beam and power, but the structure is complex, the volume is more difficult to control, and the risk of bearing and bushing wear is high. At present, only Livox has adopted this technology, and it has been mass-produced on the Xiaopeng P5.

Main manufacturer: Livox

Flash: The structure of Flash is very simple and rude. The principle is to emit a laser that can cover an area through a high-density laser source array, and use a high-sensitivity receiver to construct an image. This form of lidar has no mechanical moving parts at all, and is small in size, high in precision, and fast in scanning. However, it is easy to form side lobe interference, disperse laser energy, and affect the detection distance and resolution.

Major manufacturers: ibeo, LeddarTech, Liangdao Intelligence

OPA: The principle of OPA lidar is to form an emission array through multiple laser emission units, and to change the emission angle of the laser beam by adjusting the phase difference of each unit in the emission array, and to generate mutually reinforcing interference in the set direction to achieve A high-intensity pointing beam completes the scan.

All along, an important reason for the slow progress of OPA lidar is that the sidelobe effect is difficult to solve. The side lobe effect is a proximity effect due to optical diffraction. On the OPA lidar, the beam synthesis after the light beam passes through the OPA device is actually formed by the mutual interference of light waves, so it is easy to form array interference, so that the laser energy is dispersed, and eventually optical artifacts and other problems occur. It is for this reason that it is difficult to balance the OPA field of view and beam quality.

Major manufacturers: Quanergy, Luowei Technology, Lice Technology

MMT (Micro-Motion Technology): MMT is a unique imaging technology for LiDAR that uses a speaker-like voice coil technology with a proprietary optical array attached to the voice coil that creates micro-motion when the voice coil is energized. A micro-motion lidar can scan the image. This design has the simplest optical path, yields the highest efficiency and the lowest component count, is small in size and has virtually no loss in internal mechanical components.

In addition, compared with other lidars, MMT technology lidars emphasize resolution, especially the vertical resolution can reach 4 to 5 times that of lidars on the market. MMT technology was pioneered by lidar manufacturer Cepton, and currently only Cepton adopts this technology.

Major Manufacturer: Cepton

There are two main forms of LiDAR ranging, ToF and FMCW. ToF is also the time of flight. The distance information is calculated by directly measuring the time difference between the emitted laser and the echo signal. The detection accuracy is high and the response speed is fast. ToF is also the most widely used and mature ranging method in the market.

The FMCW coherent detection method obtains the frequency difference by coherently modulating the optical frequency of the emitted laser and coherent the echo signal with the reference light, so as to obtain the time-of-flight inversion of the target object distance. FMCW lidar has the advantages of direct measurement of velocity information and immunity to ambient light and other lidar interference. Lidar combining FMCW and OPA is considered to be the “ultimate solution” in the future.

  1. Core components

Lidar is mainly composed of four parts: transmitting module, receiving module, scanning module, signal control and processing (main control) module.

Source: Hesai Technology Prospectus

Transmitter module: side emitting EEL, vertical cavity surface emitting VCSEL, pulsed fiber laser

Receiver module: avalanche diode APD, single electron avalanche diode SPAD, silicon photomultiplier tube SiPM, PIN diode

Scanning module: MEMS micro-galvanometer, OPA silicon photonic chip,

Main control module: ADC, FPGA, SoC, etc.

The BOM cost of lidar mainly consists of three parts: hardware electronic module, optical module, and structural module. Among them, the hardware electronic module accounts for about 50%-60% of the total cost, including the respective supporting PCB boards of the transmitter and receiver, and the FPGA board is mainly responsible for the calculation, plus the main control board and power supply module; the optical module accounts for the total cost. 10%-15%, including mirrors, lenses, prisms, window glass, etc.; structural modules account for about 25% of the total cost, including supporting optical modules and some brackets inside hardware modules, including motors, bearings, brackets, housings, etc. .

3: Sorting out the LiDAR industry chain

upstream:

1. Lasers and detectors: The main upstream companies of semiconductor lasers are mainly overseas, including ams Osram, Lumentum, HAMAMATSU (Hamamatsu Photonics), II-VI, etc.; domestic suppliers include Zonghui Xinguang and Ripple Optoelectronics , Huaxin Semiconductor, Changjiang Huaxin, etc. The main suppliers of fiber lasers include overseas Luminar, Lumibird, IPG Optoelectronics, Onner, and domestic Leishen Intelligence.

In terms of detectors, overseas suppliers mainly include ams, Osram, ON Semiconductor, Firstsensor, HAMAMATSU, Sony, etc.; domestic suppliers include Lianxin Integration, Lingming Photonics, Core Vision Microelectronics, Accelerator Technology, Fushi Technology, etc.

2. FPGA: FPGA is usually used as the main control chip of lidar. The mainstream overseas suppliers are Xilinx (acquired by AMD), Altera (acquired by Intel), Lattice, Achronix, etc.; the main domestic suppliers are Ziguang Tongchuang , Zhi Polycrystalline Microelectronics, Fudan Microelectronics, Anlu Technology, Gowin Semiconductor, etc.

In terms of product performance, domestic manufacturers are still significantly behind overseas manufacturers, but the scale of logic resources and high-speed interface performance of domestic products can also meet the needs of lidar. However, in the prospectus of Hesai Technology, Hesai believes that FPGA is not the only choice for the main control chip of lidar, and high-performance MCU and DSP can also be used instead. The international mainstream suppliers of MCU include Renesas and Infineon, and the mainstream suppliers of DSP include TI and ADI.

In addition, although FPGA is the current mainstream solution, with the improvement of performance and overall system requirements, SoCs with higher integration may be widely used in future lidars, such as integrating photodetectors, front-end circuits, waveform digitization, SoC for functional modules such as waveform algorithm processing and laser pulse control. Lidar SoCs are generally developed by Lidar manufacturers themselves, and Hesai Technology is also vigorously deploying them.

3. Analog chips: mainly high-precision ADCs, which are needed in key subsystems such as lighting control, photoelectric signal conversion, and real-time processing of electrical signals. The mainstream overseas suppliers are TI, ADI, etc. The domestic Silicon Lijie, Shengbang Micro, and Chipsea Technology can supply related chips, but there is some gap in performance with the overseas leading products.

4. Optical components: including MEMS galvanometers, various optical lenses, OPA silicon optical chips, etc. MEMS galvanometers are mainly supplied by overseas manufacturers, such as HAMAMTSU, Infineon, ST, etc.; domestic Bofu Optoelectronics, Zhiwei Sensor, Yingtang Zhikong, Leishen Intelligence, etc. have related product layouts or have been applied.

In terms of optical lenses, domestic manufacturers such as Sunny Optoelectronics, Crystal Optoelectronics, Focuslight Technology, and Tengjing Technology already have mature technologies. The domestic supply chain has reached the international leading level in optical components, and has more competitive advantages in terms of cost. Substitute foreign supply chains and meet product processing needs.

OPA silicon photonics chips have not yet appeared in mass production. Overseas, Quanergy is mainly conducting research and development, and domestic Luowei Technology and Accelerator Technology are all in the layout. Limited by the slow progress of chips, OPA lidar is generally not expected to achieve mass production until 2025.

Midstream and downstream:

Valeo: It will account for 28% of the global lidar market in 2021, ranking first. As early as 2018, the Audi A8 achieved mass production of lidar for the first time. The SCALA1/2 generation that has been put on the car adopts the rotating mirror type, and the latest generation product adopts the MEMS galvanometer.

Velodyne: The pioneer of autonomous driving lidar, the current product line includes mechanical rotating and rotating mirror lidars. However, the emergence of lidar companies in the market in recent years has greatly weakened the competitiveness of their products, and their prices are relatively high compared to those of their friends. The market share in 2021 will only account for 3%.

Sagitar Juchuang: In 2021, it will account for 10% of the global lidar market and will rank second. Its product line covers mechanical rotary and MEMS galvanometers. Among them, the MEMS galvanometer lidar has received orders from many OEMs, and it has been installed on GAC Aeon LXPlus, Zhiji L7, Xiaopeng G9, Lotus Eletre, LucidAir and other models.

Livox: A subsidiary of DJI, with a market share of 7% in 2021, is currently the only lidar manufacturer that uses prismatic scanning. At present, vehicle-spec lidar products have been mass-produced on Xiaopeng P5.

Luminar: With a market share of 7% in 2021, it is the first lidar manufacturer in the market to use a 1550nm fiber laser as a light source. Using the rotating mirror + MEMS galvanometer two-dimensional scanning solution, the SAIC Viva R7, which will be launched in the second half of the year, will be equipped with the IRIS lidar using this solution.

Innoviz: The market share will be 4% in 2021. The product adopts the MEMS galvanometer scanning solution, and mainly cooperates with Tier 1. The main engine factory is currently cooperating with BMW to carry it on the BMW iX model.

Ibeo: The market share is 3% in 2021. Its products include mechanical rotary and Flash solid-state. Solid-state Flash lidar has signed a fixed-point agreement with Great Wall. In the future, a flagship model will be equipped with ibeo’s Flash solid-state lidar.

Innovusion: The market share is 3% in 2021. The technical route is the same as that of Luminar. It uses a 1550nm light source, a rotating mirror + MEMS galvanometer two-dimensional scanning solution. At present, its Falcon lidar has been mass-produced on NIO ET7 and ES7 models.

Hesai Technology: The market share will be 3% in 2021. The product line covers mechanical rotating and rotating mirror lidars. The Hesai AT128 hybrid solid-state lidar has been mass-produced on the Ideal L9.

Huawei: The market share will be 3% in 2021. At present, there is only one 96-line rotating mirror lidar, which will be mass-produced on the Polar Fox Alpha SHI version, Nezha S, and Avita 11.

Cepton: Similar to Livox, Cepton has taken a different technical route from other technology products on the market. Cepton uses MMT (MicroMotion Technology) as the scanning mechanism, which is smaller and more reliable than the hybrid solid-state lidar on the market. At present, Cepton has been bound to Tier 1 and has won large-scale orders from General Motors.

Aeva: Aeva recently announced that the first batch of AeriesII4D lidar sensors have been successfully put into production and delivered to strategic customers. AeriesII incorporates FMCW ranging technology and integrates all key sensing components including transmitters, optics and receivers on a silicon photonics chip in a compact module, making FMCW lidar easier to scale for mass production , smaller in size. But the real mass production of Aeva’s FMCW lidar won’t happen until around 2025.

Quanergy: Quanergy is the first company to enter the OPA lidar field. In May of this year, Quanergy announced that its OPA lidar achieved a detection distance of 250 meters, while the number was 100 meters 15 months ago. According to Quanergy’s official introduction, its OPA technology is based on CMOS process-compatible silicon photonic chips, which can achieve mass production. But so far, Quanergy’s S-series OPA lidars are still difficult to achieve mass production and meet customer requirements at the same time. It is said that the company’s OPA lidar is not expected to achieve mass production until after 2025.

Four: Summary

The biggest obstacles to the application of lidar in autonomous driving can be summarized as: cost, service life, and volume. At present, the reliability of lidar is mainly determined by the transceiver system and scanning system. The more mature the supply chain of the corresponding module, the lower the cost, and the easier it is to pass the vehicle certification.

Looking at the lidar market, the solution of 905nm + rotating mirror/MEMS galvanometer + ToF is the most mature at this stage, and it is also a common solution for lidar that is mainstreamed in mass-produced models. With the demand for low cost, high production capacity, and high performance, pure solid-state LiDAR may become mainstream after 2025. The mainstream view in the industry is that OPA+FMCW is the ultimate solution for future lidar, taking into account the advantages of easy mass production and high performance, and benefiting from the development of silicon photonics technology to achieve rapid cost reduction of optoelectronic devices.

Leave a Reply

Your email address will not be published.