(Producer/Author: Oriental Securities, Kuai Jian, Jiang Xueqing, Ma Tianyi)
1.1. The promotion of multiple forces is that autonomous driving is about to emerge. The two paths of autonomous driving are both different paths, and jointly promote the acceleration of high-level autonomous driving. Currently, autonomous driving is heading towards us along two paths. One is a gradual development path led by new car-making forces and traditional car companies are following, gradually realizing L3+ autonomous driving from advanced assisted driving (ADAS); the other is a technological force represented by Baidu , Google , etc., adopting a leapfrog technological development route to directly develop L4/L5 level autonomous driving/unmanned driving. Although the development paths are different, with the help of technology companies, new car manufacturers, traditional car companies and other forces, autonomous driving hardware and software technologies are becoming increasingly mature, and they jointly point to the implementation of advanced autonomous driving.
Progressive car companies generally realize L2 high-level assisted driving and move towards L3 level autonomous driving. After several years of technological accumulation, various smart car models have been released intensively and delivered to mass production, and have begun to enter the harvest period. Entering 2021, NIO ET7, ET5, and Xiaopeng P5, which have L3 level automatic driving capabilities have been released one after another. Huawei Alpha S, which meets the L2-L4 level, was also released in April 21, and the industry has begun to enter the era of L3+ autonomous driving.
The global autonomous driving market is expected to usher in a period of rapid growth. According to IDC, the global shipments of autonomous passenger cars with autonomous vehicles are expected to grow from 27.74 million in 2020 to 54.25 million in 2024, of which the shipments of autonomous vehicles of L3 and above have increased from 32,000 in 2020 to 863,000 in 24, with a CAGR of up to 128% in 2020.
And on the other hand, autonomous driving technology is becoming more and more mature. Up to now, a total of seven companies have obtained California full driverless licenses, including Chinese-based Antu, Baidu, Wenyuanzhixing and American-based GM Cruise, Nuro, Waymo, and Zoox (under Amazon ). Waymo and GM Cruise lead with extremely high total mileage and average mileage per takeover; Antu (AutoX), Wenyuan Zhixing (WeRide), etc. followed closely. Comparing the test data in 2019 and 2020, the number of car takeovers in each company has decreased significantly, and the average total mileage between each takeover has increased significantly. (Report source: Future Think Tank)
automatic driving commercialization has begun. After years of accumulation of unmanned driving technology, autonomous driving has begun to be first implemented in scenarios such as unmanned taxi travel services, unmanned delivery vehicles and long-distance cargo transportation. In China, Baidu, Antu, Didi , Wenyuan Zhixing, Pony Smart Travel and other companies have launched driverless taxi services and entered the trial operation stage. Among them, Baidu Apollo driverless Robotaxi has become the first batch of "shared unmanned vehicles" in China, officially starting normalized commercial operations and fully opening them to the public. Overseas, Waymo Robotaxi started commercial operations as early as 18 years ago, and the fleet size continued to expand.
Although autonomous driving is still in its initial commercialization stage and is still a distance from large-scale commercialization, as Robotaxi gradually moves from limited testing scenarios to more diversified operation scenarios, from having safety officers ready to take over to completely unmanned driving, from free experience to paid operations, the application ecosystem of autonomous driving in Robotaxi and other scenarios will become increasingly mature.
1.2 The general trend of multi-sensor fusion is that lidar is indispensable
camera technology is mature and has low cost, becoming the first perception hardware to load and consume the largest amount. car camera is the main vision sensor of the ADAS system and is also one of the most mature car sensors. However, since the camera passively receives visible light like the human eye, the visual effect is poor in the case of backlight or complex light and shadow, and is easily affected by bad weather.
mm wave radar is least affected by the weather environment and has the best performance all-weather.The working principle of millimeter-wave radar and lidar is similar. The frequency bands commonly used in the automotive field are 24GHz, 77GHz and 79GHz, which correspond to short, long and medium-distance radar , respectively. Due to its long wavelength and good circumference ability, millimeter-wave radar is least affected by the weather environment. However, due to its long wavelength, the detection accuracy is greatly reduced. Lidar has the best accuracy and meets the needs of L3-L5 autonomous driving.
lidar uses laser as carrier wave, and its wavelength is shorter than millimeter wave, so it has high detection accuracy and long distance. LiDAR can also recycle information from laser ruler in different directions, use points to form lines and lines to form surfaces, forming obstacles 3D " point cloud " image. Due to the difficulty of technology and high cost, large-scale loading has not yet been achieved. With the increasing maturity of the industrial chain in the future, the lidar industry may usher in an explosion after the cost drops.
is the general trend of multi-sensor fusion, and lidar is indispensable. In the field of autonomous driving perception technology, there are currently two major camps, "visual perception" represented by Tesla and "3D lidar fusion perception" solutions represented by Waymo, NIO , Xiaopeng , etc. The former is "light perception and focused on algorithms", using low-cost cameras for environmental perception, supplemented by high-performance computing, which requires high computing power for vision-based neural network algorithms; the latter mainly relies on lidar to create a 3D environment map of surrounding environment perception, forming a fusion redundant perception solution of "camera + millimeter wave radar + lidar".
Although pure vision solutions have certain cost advantages and can meet the current L2-level ADAS perception needs, with the gradual evolution of autonomous driving, the amount of data in the perception layer has increased exponentially. Weak perception will put higher requirements on the performance and computing power of the chip, increasing the difficulty of implementation. In addition, the lack of performance and recognition accuracy of the camera itself (such as not being able to directly give distance, reducing the three-dimensional world to two-dimensional imaging, being subject to weather, etc.) also restricts the development and popularization of pure visual perception solutions in high-level autonomous driving. We believe that in order to achieve comprehensive coverage of unmanned driving functionality and safety, sensor fusion and redundancy will become the main theme of the future. As a key link in which detection accuracy and resolution is higher, the continuous iteration and maturity of its technical processes and the gradual decline in costs will also promote its large-scale vehicle loading application in L3 and above models.
1.3 The first lidar mass-produced vehicle was launched, opening the prelude to large-scale loading
Robotaxi was the main battlefield of lidar before. Compared with the passenger car field, Robotaxi is positioned at the L4-L5 level unmanned driving, so it has the highest requirements for sensing layer detection performance. At the same time, since the unmanned driving companies of the vehicle owner are often large automobile/technology manufacturers with strong capital strength, the price of lidar and the integration with the vehicle body are relatively low. Robotaxi of many operators has been equipped with lidar, which is the main battlefield of lidar before.
passenger cars ushered in a small climax of lidar loading. Four cars including NIO ET7, ET5, Xiaopeng P5, and L2-L4-level Huawei Alpha S, which have L3-level autonomous driving capabilities, have been released one after another, all equipped with lidars. Among them, Xiaopeng P5 is the world's first mass-produced car equipped with laser radar. As "gradual" smart car companies gradually achieve the leap from L2 assisted driving to L3 autonomous driving, the demand for lidar loading has reached a climax and will enter the stage of centralized loading, and passenger cars will also become the main battlefield of lidar in the future.
"gradual" and "leapfrog" dual-wheel drive, lidar starts centralized loading, and the market size is growing rapidly. On the one hand, with the gradual evolution of autonomous driving, the integration and redundancy of sensors have become the key to liberating drivers' hands and feet and ensuring their safety. Lidar will become an indispensable perception hardware; on the other hand, with the commercialization of Robotaxi/Roboruck, the fleet scale in this field will accelerate in the future. Sullivan Research predicts that the scale of new landing fleets will exceed 600,000 by 2025, bringing broad downstream space to the application of lidar. The two will jointly drive the lidar market to usher in prosperity.Sullivan Research predicts that the global lidar market size will be US$13.5 billion by 25 years, with a CAGR of 65% of 19-25E; among which the market sizes in the unmanned driving and ADAS fields will increase to US$3.5/4.6 billion respectively, with a CAGR of 19-25E to 81%/84%, which will account for about 60% of the total lidar scale.
2.1 Lidar types are diverse, and the technological development is changing with each passing day
Lidar is a sensor that uses laser to achieve accurate ranging. In a broad sense, it can be regarded as a camera with 3D depth information and is known as the "eye of a robot." Since its birth, the lidar industry has kept up with the cutting-edge development of underlying devices and has shown outstanding characteristics of high technical level. From multi-line scanning lidar that has been widely recognized in autonomous driving technology, to solid-state lidar and FMCW lidar with continuous innovation in technical solutions, and in recent years, lidar has always been a representative of the development and application of emerging technologies. The lidar structure is precise and complex, and is mainly composed of four core components: laser system, reception system, signal processing unit and scanning module.
laser illuminates the emitted laser light in a pulsed manner, and then illuminates the obstacle and scans the object 3D. The reflected light converges on the receiver through the lens group. The signal processing unit is responsible for controlling the transmission of the laser, converting the received analog signal into a digital signal, and finally entering the main control chip for data processing and calculation.
Since each functional module of lidar has a variety of technical implementation methods, there are many types of lidar under different combinations based on each classification, and the technical route is in the stage of rapid development and iteration.
According to the ranging method: lidar can be divided into time of flight (Time of Flight, ToF) ranging method, frequency modulated continuous wave (FMCW) ranging method based on coherent detection, and triangular ranging method. Among them, ToF and FMCW can achieve a longer range (100~250m) under outdoor sunlight, which is the preferred solution for vehicle-mounted lidar . ToF is the mainstream solution for in-vehicle medium and long-distance lidar in the market. In the future, with the maturity of the FMCW lidar whole machine and the upstream industrial chain, ToF and FMCW lidar may coexist in the market.
According to the scanning method: lidar can be divided into mechanical lidar with overall rotation, semi-solid-state lidar with transceiver module stationary semi-solid-state lidar and solid-state lidar. The difference is whether there are active components. 1) Mechanical lidar: laser scanning is realized through mechanical rotation, and continuously rotates under the motor drive. The laser beam in the vertical plane changes from "line" to "surface" and then forms multiple laser "surfaces", thereby achieving 360° 3D scanning in the detection area. 2) Semi-solid state solution: It includes three types: MEMS, mirror-type, and prism-type. Its characteristics are that the transceiver unit is decoupled from the scanning component and the transceiver unit no longer performs mechanical movement, and is small in size and low in cost. It is currently the mainstream choice. 3) Solid-state lidar: It mainly includes two implementation methods: optical phased array (OPA) and flash (Flash) type. It completely eliminates the mechanical scanning structure and does not have any moving parts inside. Laser scanning in horizontal and vertical directions is achieved electronically, greatly reducing laser transceiver devices, thereby reducing costs. The miniaturized structure also improves performance stability. It is expected to dominate the future with better cost performance.
2.2 Scan module: Semi-solid state, solid state comes from behind
For vehicle manufacturers or planners, automotive grade, mass production, and controllable cost are the main considerations for large-scale loading of lidars. laser radar technology routes vary greatly and have low homology. At present, vehicle-mounted lidar is constantly iterating along the development path of mechanical-semi-solid-solid state. Due to its high cost, mechanical lidar is more suitable for R&D and testing projects such as unmanned driving. It is still to be tested in the fields of mass-produced vehicles and passenger cars. In the short term, semi-solid state is expected to be the first to be on the vehicle and dominate medium and long-distance lidar. In the long run, when the solid-state Flash and OPA with the highest technical process level mature, they may become the mainstream technical direction.
2.2.1 Mechanical lidar: High precision is accompanied by high cost and difficult to mass production bottleneck
Performance: Mechanical lidar is the earliest developed and most mature product. It has become the first choice for sensors for driverless projects with its advantages such as simple principles, easy drive, and easy to achieve level 360° scanning.
Cost: high cost and little room for price reduction. The internal structure of mechanical lidar is precise, with many parts and complex assembly process. Especially for high-wire harness lidar, the more wire harnesses, the number of transmitting and receiving modules needs to be increased accordingly, and the later maintenance costs are superimposed, resulting in high costs, and the cost of high-wire harness lidar is difficult to be less than US$3,000.
Automobile specifications: It is difficult to meet the automotive specifications. High-frequency rotation and complex mechanical structures make their rotating parts easily wear, which affects detection accuracy. Currently, the average failure time of most products is only 1000-3000 hours, which is difficult to meet the minimum requirement of 13000 hours for automotive-grade equipment.
mass production: Mechanical type is an option for cost-insensitive Robotaxi/unmanned driving companies, but for vehicle manufacturers and solution providers, large-scale mass production is difficult. Velodyne, the leading company, launched a 64-line mechanical lidar product in 2007. In 2010, Google's first self-driving car used the Velodyne lidar solution. Now the product solution is widely used in test models such as Baidu and Uber.
2.2.2 Semi-solid state lidar: Be the first to get on the car, it is the current choice
Semi-solid state lidar transceiver module is stationary, and only the scanner is mechanically moved. The rotating mirror, MEMS micro galvanometer, etc. replace the rotating scanner in the previous mechanical type. It is smaller in size, higher integration and lower cost, suitable for the needs of mass-produced models of front-loading, and is the choice of some mainstream car manufacturers at present.
semi-solid state reversing mirror type: One-dimensional reversing mirror technology develops in two-dimensional and becomes the mainstream lidar product
Performance: The reversing mirror lidar power consumption is relatively low and the heat dissipation is low, so the reliability is high. The disadvantage is that it is difficult to integrate to further reduce costs, and the number of scanning lines of the one-dimensional reversing mirror is small, and the scanning angle cannot reach 360°. The function of few transmitters and multiple wire harnesses is realized through the two-dimensional reversing mirror.
Cost/price: Take Innovusion’s image-grade ultra-long-range lidar Falcon equipped with NIO ET7 as an example, its price after mass production is about US$500-1,000.
Automotive Regulations & Mass Production: Valeo's first generation SCALA mass-produced in 2017 is the world's first lidar to pass the automotive grade certification. It was first launched on Audi A8 in the same year, so it became the first technical solution to pass the automotive specifications, cost controllable, can meet the performance requirements of car companies, and achieve mass supply. The current shipment of the SCALA series has exceeded 150,000, and its customers include BMW , Mercedes-Benz , etc.; in October 2020, Radius Intelligence CH32 became the second lidar to obtain automotive specification certification. In 21 years, the 96-line lidar developed by Huawei has also been the first to be installed in the Alpha S Huawei HI version of the Extreme Fox. (Report source: Future Think Tank)
Semi-solid MEMS micro galvanometer: It has met the requirements of automotive regulations, and the landing process is accelerated.
uses semiconductor "micro-moving" devices instead of macro-mechanical scanners as laser beam scanning elements. By controlling the tiny mirror plane and torsional reciprocating motion, the emission angle of a single emitter is changed to scan, forming a wide scanning angle and a large scanning range, thereby forming a high-density point cloud map at an ultra-high scanning speed.
Performance: Small size and higher integration. MEMS micro galvanometer helps the lidar get rid of bulky motors, polyprisms and other mechanical motion devices. The millimeter-level micro galvanometer greatly reduces the size of the lidar and greatly simplifies the scanner structure, making it have the advantages of high performance, stability and reliability, and easy to manufacture. However, MEMS micro-galvanometer lidar will have problems such as low signal-to-noise ratio and short effective distance.
Cost/price: Reduce the number of lasers and detectors, reduce costs. The number of wire harnesses required for traditional mechanical lidars to realize, the number of transmission modules and reception modules is required.The micro-galvanometer accurately controls the deflection angle, and achieves the equivalent mechanical laser radar coverage area and point cloud density by controlling the scanning path, greatly reducing costs. MEMS micro galvanometers have been commercially used in the field of projection display for many years, and the upstream supply chain is relatively mature. Luminar's MEMS semi-solid-state lidar has reduced the manufacturing cost to US$500-1,000, making large-scale mass production possible. Radius Intelligence MEMS LiDAR LS20B Series 20 years priced at $999/$1299.
Automobile regulations & mass production: Domestic brands are the first to be implemented. MEMS lidar can take into account the needs of automotive mass production and high performance. The MEMS lidar RS-LiDAR-M1 of Sagitar Juchuang was shipped in batches in December 2020, becoming the world's first automotive-grade MEMS lidar delivered in bulk, and has established cooperation with 11 car companies including GAC Aion, WM Motor, Zekr , etc. Overseas, Luminar already has more than 50 industry partners around the world, including , Volvo , SAIC Feifan Automobile, Pony Smart Strip, etc.
Semi-solid prism: Livox is unique, self-built production capacity is bound to Xiaopeng
adopts a double wedge prism structure. The laser deflects twice after passing through the two wedge prisms. As long as the relative speed of the prisms on both sides can be controlled, the scanning form of the laser beam can be controlled.
Performance: High point cloud density and long detection distance. The scanning trajectory of the prism lidar is petal-shaped. When the scanning speed is properly controlled, long-term scanning at the same position can cover almost the entire area, thus avoiding multiple calibrations of traditional rotating lidars.
Price: Livox Horizon official website priced at RMB 7199.
Automotive Regulations & Mass Production: Livox has provided automotive-grade lidar technology for new models mass-produced by Xiaopeng since 2021. Livox Horiz (Horizon customized version) began mass production in its own automotive-grade manufacturing center in 2021, with an average annual production capacity of 100,000 units, and can expand the line within 3 months based on the growth demand of front-load mass-produced customers. The new product HAP has established a partnership with Xiaopeng Motors and FAW Auto and was mass-produced on 21Q4.
2.2.3 Solid-state lidar: The technology is about to mature and is the choice for the future
From the perspective of performance, there are no moving parts inside the solid-state lidar, and the reliability has been greatly improved. At the same time, because components can be IC-based and highly automated assembly and adjustment, it is easier to achieve mass production and significantly reduce mass production costs, so it is also recognized as the mainstream evolution route for lidar to achieve large-scale vehicle-mounted vehicle. From a technical perspective, solid-state lidar has high requirements for parts and is difficult to integrate manufacturing processes. At present, the two technical routes, Flash and OPA, are not yet mature, and they still need to match high-power lasers, high-performance photodetectors, and other components, which is still a gap from the actual landing.
pure solid state Flash: Global flash imaging, achieving the world's first mass production cooperation on solid state lidar vehicle regulations
Flash Lidar refers to a lidar that uses the flash principle to achieve a flash ( laser pulse ) imaging. It uses a vertical cavity surface transmitter (VCSEL) at the emission end to emit a large area array laser covering the detection area in a short time, and then uses a highly sensitive receiver to complete the drawing of the surrounding image.
Performance: Instantaneous recording, high integration, but close detection distance and low resolution. The advantage of Flash lidar is that it has no scanners, which can quickly record the entire scene, avoiding errors caused by the movement of the target or the lidar itself during the scanning process, and has high integration, small size, and a chip-level process, suitable for mass production. However, the power density of Flash lidar is low, which makes it difficult for its effective distance to exceed 50 meters and its resolution is relatively low.
Cost: Currently, Flash technology is not mature, the price is correspondingly high, and there is room for downward in the future. Take Ouster's 2020 Flash LiDAR ES2 as an example. The initial price for automotive-based mass production projects will be $600, with the goal of reducing it to $100, and it is planned to achieve mass production in 2024.
Automotive specifications: small size, high stability, no mechanical components, and it is expected to meet the requirements of high-grade automotive specifications after the technology is mature.
mass production: Due to the limitation of detection distance, the Flash solution is mainly used in low-speed autonomous driving solutions, such as unmanned takeaway vehicles, unmanned logistics vehicles, etc., which have lower requirements for detection distances. The motorcycle models under the Great Wall Motor WEY brand series are confirmed to be equipped with IBEO's Flash lidar IbeoNEXT, becoming the world's first automotive-based mass production cooperation for pure solid-state lidar.
Pure solid state OPA: high system integration optical phased array technology
OPA LiDAR uses the coherence principle (similar to the fact that two circles of water waves superimpose each other, some directions will cancel each other, and some will enhance each other), and uses multiple light sources to form an array. By adjusting the phase difference of each emission unit in the emission array, the laser exit angle is changed, and the time difference of emission of each light source is controlled, so as to synthesize a flexible, precise and controllable main beam to realize scanning of different directions.
Performance: small size, fast scanning speed, high accuracy. OPA realizes a beam scanning without any mechanical components, and the regulation speed can be very fast, easily reaching point scanning speeds of MHz or even GHz, while also having very little power consumption. Secondly, OPA adopts an array grating transceiver structure to avoid the late alignment process. It can also use semiconductor technology to achieve the integration of the detection system, further reducing the volume and reducing costs.
Technology development: Problems such as processing technology, scanning angle, and distance are still to be broken. The beam synthesis of the light beam after passing through the optical phased array device is actually formed by the mutual interference of light waves, which easily forms array interference, causing the laser energy to be dispersed (sidelobe effect). Currently, OPA laser lightning is not mature in reducing side lobe effects, processing technology, detection distance and other technical difficulties.
Cost: The current OPA technical level barrier is high, and there is a lot of room for cost reduction in the future. The early R&D costs of OPA lidar are relatively high, and the future technology will mature, which will drive the price of lidar products to decline. At the 2016 CES exhibition, Quanergy released the "world's first OPA solid-state lidar" S3, with a size of only 90*60*60mm, with a small size, low power consumption and low cost, and a cost of only US$200 per unit, but the company has not yet mass-produced it.
Automotive specifications: small size, high stability, no mechanical components, and it is expected to meet the requirements of high-grade automotive specifications after the technology is mature.
mass production: The technology is not yet mature, and mass production will still take time. Although OPA technology is quite advanced, the size and accuracy of related components are very high. For example, the wavelength of the lidar is about 1 micron. In order to weaken the impact of side lobes, the spacing between adjacent cells of the array needs to be less than 500nm, which is difficult to integrate and manufacture. It is expected that it will take about 5 years to truly land.
2.3 Transceiver module: Core hardware integration and chip architecture are necessary to planarize the emission end of
, and emit from edge to vertical emission
semiconductor transmitters can be divided into EEL (edge emission laser) and VCSEL ( vertical cavity surface emission laser ). Previously, VCSEL lasers had the defect of low luminous density and power, which could only be applied to scenarios where distance measurement requirements were close (usually 50 m). In recent years, with the development of multi-layer junction VCSEL lasers, their luminous power density has been increased by 5 to 10 times, which provides the possibility for the use of VCSEL to develop long-distance lidars. Combining the production costs and product reliability benefits brought by its planarization, VCSEL is expected to gradually replace EEL in the future. The
transmitter gradually adopts planar laser devices. As a detection light source, EEL has the advantage of high luminous power density, but its luminous surface is located on the side of the semiconductor wafer. During use, the process steps of cutting, flip, coating, and re-cut are often only integrated with the circuit board through a single one-on-mount method. Moreover, each laser needs to use separate optical devices to compress the beam-draining divergent angle and independently manual installation and adjustment, which greatly relies on the manual installation and adjustment technology of production line workers. The production cost is high and the consistency is difficult to guarantee.The VCSEL light emitting surface is parallel to the semiconductor wafer and has the characteristics of luminescence on the surface. The laser array formed by it is easy to bond with planar circuit chips. It is guaranteed by semiconductor processing equipment at the accuracy level, and there is no need to separate assembly and adjustment of each laser. It is easy to integrate with the silicon material microlens processed on the surface to improve the beam quality.
The single photon detector using CMOS technology
APD (avalanche photodiode) and SPAD (single photon avalanche diode) are two photodetectors that convert light energy into current. The SPAD arraying process is mature and is expected to replace APD. Compared with APD, SPAD has single-photon detection capability and is widely used in the fluorescence detection field of biomedical and the field of nuclear magnetic imaging. However, because the output signal amplitude of SPAD is the same, it cannot measure the light intensity, while APD outputs an analog signal, which can obtain the target's grayscale information and has a large dynamic range, resulting in the measurement sensitivity of SPAD at the receiving end of the lidar is not as popular as that of currently used in lidar.
In recent years, many detector companies at home and abroad have continuously optimized the quantum efficiency of single-photon devices in the near-infrared band, and have gradually surpassed APD in actual detection sensitivity. In addition, SPAD's CMOS arraying process is relatively mature and easy to configure, while APD is difficult to implement arraying due to the need for specialized technologies. In the future, with the further optimization of design and process, the advantages of SPAD on APD performance will become more and more obvious.
Hardware integration and chip architecture are imperative, and the key to reducing costs and increasing efficiency lies in it. The core lasers, detectors, control and processing units in lidar system can all benefit from the development of the semiconductor industry. At present, lidar still has problems such as many parts, high production costs and low reliability. Transmitter and reception unit arraying and core module chipization are future development trends. Lidar with chip architecture can integrate hundreds of discrete devices into one chip, reducing material costs while saving labor production costs of independent optical installation and adjustment of each laser. In addition, the reduction in the number of devices can significantly reduce the probability of system failure due to single device failure and improve reliability. Therefore, the integration and chipization of core hardware and modules are the key to miniaturizing, lightweighting and meeting automotive specification requirements, which will bring qualitative changes to the cost reduction and efficiency of lidar and large-scale applications.
is customized to develop VCESL dedicated analog digital chips at the transmitting end. The VCSEL multi-channel driver chip can provide peak current of tens of amperes and narrow pulse width driving capability in nanoseconds to meet the needs of lidar detection. Moreover, in the future, through the integration of VCSEL array and driver chip packaging level, the parasitic inductance of the drive loop can be further reduced, narrower pulse width and higher electro-optical conversion efficiency, thereby further improving the distance measurement accuracy and remote measurement capability of the lidar.
realizes a monolithic integrated SoC of detector and circuit function modules under the CMOS process at the receiving end and the information processing unit. The single-photon receiver has an integrated SoC chip on chip. Through the integrated detector, front-end circuit, algorithm processing circuit, laser pulse control and other modules, it can directly output distance and reflectivity information, gradually replacing the function of the main control chip FPGA. In the future, with the continuous increase in the scale of line columns and surface arrays, the CMOS process nodes will be gradually upgraded. The single-photon receiver SoC will achieve stronger computing power, lower power consumption and higher integration.
upstream and downstream resonance, and the lidar industry chain is maturing. Upstream manufacturers of the lidar industry chain are responsible for providing optoelectronic components required for laser emission, laser scanning, laser reception and information processing. They are integrated and produced by midstream manufacturers, and finally applied to multiple fields such as autonomous driving, ADAS, map surveys, etc., to form a complete industrial chain. On the demand side, downstream industries are ready to go. The rapid rise of autonomous driving has brought new opportunities to the development of the lidar industry; on the supply side, the technology of mid- and upstream enterprises is constantly developing.The technical processes of upstream optoelectronic devices suppliers are constantly iterating and upgrading, and the rapid iteration of the technical paths of midstream lidar enterprises drive the increasingly mature industrial chain, which has also promoted the accelerated implementation of lidar products. (Report source: Future Think Tank)
3.1 Upstream determines product performance, while overseas manufacturers lead domestic manufacturers follow
upstream determines product performance and cost. Lidar is essentially an optoelectronic system composed of a variety of optoelectronic components. The optical system consisting of optical path design, laser, detector, scanning module, optical components, optical drive chips and main control chips accounts for more than 70% of the cost of the lidar machine, so it is closely related to the detection performance, cost and reliability of the lidar. In addition, most of the reason for the high cost of lidar is the calibration work. During the production and manufacturing process of multi-wire-harness lidar, multiple transmitting and receiving circuit boards need to be installed into a precision-structured metal shell. At the same time, during the debugging process, workers need to debug the transmission and reception of each laser to ensure its distance measurement accuracy at different test distances. Therefore, automated, high-precision detection and traceability equipment are also crucial.
Overseas manufacturers are generally leading, domestic companies are showing their strengths, and domestic manufacturers of optical components have advantages. Upstream core component manufacturers such as precision instruments and chips are currently basically monopolized by major foreign manufacturers, and are ahead of domestic manufacturers in terms of technology and customer base. However, driven by the dual driving of policies and downstream market environment, domestic manufacturers have been catching up in recent years and have made breakthroughs in sub-sectors.
In the chip field, FPGA is mainly led by three overseas manufacturers, Xilinx, Intel's Altera and Lattice. The main domestic suppliers include Unigroup Guoxin, etc. Analog chip suppliers are led by ADI and Texas Instruments (TI). Domestic manufacturers such as China Resources Micro and Shengbang Microelectronics are actively making arrangements to catch up with the richness of automotive-grade products and technical level. Lasers and detectors are important components of lidar and often need to meet the customized needs of different technical routes.
laser is led by European and American companies such as AMS and Lumentum. The main detector companies include Hamamatsu Photonics, ON Semi, Sony, etc. In my country, Ripple Optoelectronics (laser), Core Vision (scanner), Lingming Photonics (detector) and other companies have begun to emerge and develop rapidly. The product performance has basically approached the level of foreign supply chains, and domestic lasers and detectors that have passed the automotive certification (AEC-Q102) have appeared, and they have more customized flexibility. In terms of optical components, lidar optical components are mainly independently developed and designed by lidar companies, and optical parts manufacturing companies are selected to complete the production and processing processes, or head optical companies participate in the relevant design. The automotive standardization of optical components of
is the basis for the realization of automotive-grade lidar. At present, the technical level of the domestic optical component supply chain has completely reached or exceeded the level of foreign supply chains. Moreover, my country's manufacturing companies have natural advantages of being close to the downstream market and are more competitive in terms of costs. They can completely replace foreign supply chains and meet product processing needs. Leading companies with deep domestic optical technology precipitation such as Sunny Optics, Crystal Optoelectronics, Yongxin Optics, etc. are expected to benefit deeply in the long run.
3.2 Competition in the middle reaches intensified, and the domestic market is full of flowers
competition in the middle reaches intensified, and participating in the diversification of enterprises. With the vigorous development of the lidar industry, the track has become more prosperous, and new participants and their technical routes have become more diverse. Overseas brands such as Velodyne, IBEO, Quanergy, etc. include well-known start-ups such as Hesai Technology, Sagitar Juchuang, and RadiShen Intelligence. There are also technology companies such as Huawei and DJI entering the market across the border, and the domestic market is showing a trend of blooming.
Foreign companies are ahead of the listing boom. The foreign lidar industry started early, including veteran manufacturers Velodyne, Valeo, IBEO and rising stars Luminar, Ouster, Innoviz, etc.After Velodyne launched the 360° rotary 64-line lidar in 2005, it has become the world's leading lidar supplier. Its products are widely used by leading driverless companies such as Google and Baidu. It once accounted for more than 80% of the world's lidar orders. In recent years, it has also actively deployed hybrid solid-state and solid-state lidars to promote the mass production of products. Valeo is one of the world's largest suppliers of auto parts, receiving orders worth approximately 500 million euros from four mainstream global automakers in 2019. The third generation of SCALA lidar was released at the recent CES show, which is expected to be available in 24 years. Different from the previous two generations of micro-revolving mirror solutions, SCALA 3 has begun to adopt MEMS technology. Over the past 20 years, six well-known overseas lidar companies including Velodyne and Luminar have been listed through SPAC mergers, and Quanergy and Cepton are also preparing to go public, marking that the overseas lidar industry is expected to enter a more mature stage.
Domestic lidar companies are emerging. According to Yole statistics, in the global automotive lidar market in 21 years, domestic companies Sagitar Juchuang / Livox / Huawei / Hesai Technology / Tudaten have a place with 10%/7%/3%/3%/3% respectively. Among them, Sagitar Juchuang and Livox ranked 2/4 in the world and belonged to the first echelon in China. Hesai Technology and Sagitar Juchuang mainly focus on mechanical lidar. While successfully seizing part of Velodyne's market share, they have also begun to actively deploy semi-solid-state lidar paths. Huawei and DJI have entered the transboundary mirror/prism-type semi-solid solution, and the products have been successfully mass-produced. The addition of major technology manufacturers will also help my country improve its technical level and enrich its technical route. In the field of solid-state lidar, my country also has startups with strong R&D capabilities, such as Beixing Photonics (Flash), Luowei Technology (OPA), Guoke Optical Core (FMCW), etc., and all forces are blooming to jointly promote the continued prosperity of my country's lidar industry and narrow the gap with foreign countries.
(This article is for reference only and does not represent any of our investment advice. If you need to use relevant information, please refer to the original text of the report.)
selected report source: [Future Think Tank]. Future Think Tank - Official Website
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-