Lidar is a measurement tool that measures the distance to a target by illuminating a pulsed laser and measuring the return time of the reflected pulse with a sensor. Its core advantage is to use the high-frequency characteristics of lasers to measure a large number of high-speed position and velocity information to form accurate and clear 3D modeling of objects. Velodyne brought lidar to autonomous driving for the first time by applying lidar to the DARPA Driverless Car Challenge, and the industry has since grown significantly as downstream applications such as ADAS continue to grow. According to Frost & Sullivan data, ADAS and unmanned driving will become the main downstream applications by 2025, accounting for 34.64% and 26.30% of the lidar market, respectively.
In this issue of intelligent internal reference, we recommend Huachuang Securities' report "Lidar: Automotive Intelligent Accelerated Penetration, LiDAR Track is Expected to Usher in a Ramp-up Period", which comprehensively discusses the development and future trend of the lidar industry.
Source: Huachuang Securities
Original title:
"LiDAR: Automotive Intelligent Accelerated Penetration, LiDAR Track is Expected to Usher in a Volume Period"
Author: Geng Chen Yue Yang Gao Yuan
First, Lidar, the automotive field has good prospects
LiDAR (LiDAR, Laser Detecting and Ranging) is a measurement tool that measures the distance to a target by irradiating a pulsed laser and measuring the return time of the reflected pulse with a sensor. Its working principle is to send a detection signal (laser beam) to the target, and then compare the signal received from the target (target echo) with the transmitted signal, and after appropriate processing, the relevant information of the target can be obtained, so as to detect, track and identify the surrounding environment. It consists of a laser transmitter, an optical receiver, a turntable and an information processing system. Its core advantage is to use the high-frequency characteristics of lasers to measure a large number of high-speed position and velocity information to form accurate and clear 3D modeling of objects.
Lidar was introduced into the automotive field at the beginning of the 21st century and ushered in rapid development with the increase in ADAS penetration. Lidar was first used in the field of map mapping, and the high precision requirements made the cost of lidar high. Velodyne brings lidar to autonomous driving for the first time in the DARPA Driverless Car Challenge. Subsequently, with the continuous development of downstream applications such as ADAS, the number of enterprises in the field of lidar continues to increase, and with the continuous development of lidar, the product performance of lidar has steadily improved, the cost has dropped significantly, and the industry has ushered in great development.
LiDAR industry development history
LiDAR products can be evaluated and compared from the aspects of dominant parameters, measured performance and implicit indicators. Dominant parameters mainly refer to the information listed in the product parameter table, mainly including telemetry capability, point frequency, angular resolution, field of view, accuracy, power consumption and integration. The measured performance mainly refers to the product performance measured in the actual use of LiDAR, which determines the effective sensing distance of driverless cars and service robots to the surrounding environment. Compared with dominant parameters, users will pay more attention to the measured performance, but LiDAR, as an emerging product that has only gained high attention in the market in recent years, has limited public test data that can be referenced. Implicit indicators include the reliability, safety, service life, cost control, and mass production of lidar products, which are more difficult to quantify and lack public information.
The main parameters of lidar correspond to the performance description
The technical route of lidar has four main dimensions: ranging principle, light source, detector, and beam manipulation. LiDAR mainly includes four major systems: laser emission, scanning system, laser reception and information processing, and the four systems complement each other. According to the different characteristics of these four systems, the lidar technology route can be elaborated from four different dimensions. Among them, the light source and detector are the transmitting and receiving ends of LiDAR, beam manipulation is the scanning method of LiDAR, and ranging provides distance information for information processing. Lidar can be classified according to four main dimensions, and each different classification method can be further subdivided into different technical routes, and there are great differences between different routes.
Lidar classification
According to the classification of ranging methods, lidar can be divided into 4 types. There are four main categories of lidar according to the principle of ranging: Time-of-flight method (ToF, Time of Flight), FMCW (FMCW, Frequency Modulated Continuous Wave), triangulation ranging method and phase method. The two main measurement methods are ToF and FMCW. The ToF measurement principle is to measure the distance by recording the time between the emission of a short pulse and the receipt of the reflected light, and measure the position of the object by the angle of the reflected light during the measurement. The measurement principle of FMCW is to linearly modulate the optical frequency of the emitted laser, so that the echo signal and the reference light are coherent to obtain the frequency difference to indirectly obtain the time-of-flight reverse target distance, the advantage is strong anti-interference, and the speed can be directly measured.
The main types of ranging methods
In the lidar ranging method, ToF and FMCW can achieve long range measurement (100~250m) under outdoor sunlight, which is the preferred scheme for vehicle-mounted lidar. ToF is currently the mainstream scheme of the vehicle-mounted medium and long-distance lidar market, with a very high laser emission frequency and high-precision detection advantages, but the maximum laser power of ToF lidar is limited, there is a bottleneck in the detection distance, and it will be interfered by sunlight during the day and generate noise in the process of receiving signals. In addition to the high cost, FMCW lidar has the advantages of direct measurement of speed information, anti-interference, high distance, and in the future, with the maturity of FMCW lidar and the upstream industry chain, its proportion is expected to increase, becoming the main ranging method coexisting with ToF.
ToF versus FMCW ranging methods
The detector in LiDAR, that is, the photodetector, can be divided into four categories: PIN PD, APD, SPAD, SiPM, and APD is the current mainstream.
PIN PD (PIN PHOTODIODE) is suitable for FMCW ranging lidar with low cost;
APD (Avalanche Photo Diode), that is, avalanche photodiode, the more mature APD is widely used in ToF lidar, is currently the most widely used photodetector device;
SPAD (single-photon avalanche diode) has long-distance detection ability at low laser power, but the disadvantage is that too sensitive reception performance will bring problems such as channel crosstalk and parasitic pulse, and its circuit design and other process problems also bring high manufacturing costs;
SiPM (Silicon Photomultiplier Tube) is an array of multiple SPADs that allow for higher detectable range and use with array light sources, making it easier to integrate CMOS technology.
Comparison of optical semiconductor detectors that can be used for lidar
The EEL manufacturing process is complex, and VCSEL is expected to usher in rapid development in the future. In terms of laser light source, it can be divided into two categories from the emission dimension: side emission (EEL) and vertical cavity surface emission (VCSEL). According to the prospectus of Hesai Technology, EEL as a detection light source has the advantage of high luminous power density, but because its light-emitting surface is located on the side of the semiconductor wafer, the complex process steps of cutting, flipping, coating and re-cutting are required during use, and each laser greatly relies on the manual assembly and adjustment technology of production line workers, and the production cost is high and consistency is difficult to guarantee. VCSEL because the light-emitting surface is parallel to the semiconductor wafer, the laser array formed by it is easy to bond with the planar circuit chip, no need to arrange each laser separately, and it is easy to integrate with the silicon material microlens of the on-plane process, which can effectively improve the beam quality. In recent years, many VCSEL laser companies at home and abroad have developed multi-junction VCSEL lasers, which has effectively improved the optical power density of VCSEL and enabled VCSEL to be used in the field of long-distance lidar. From the perspective of production cost and product performance reliability, VCSEL is expected to gradually replace EEL in the future.
905nm is a near-infrared laser, which is easily absorbed by the human retina and causes retinal damage, so the 905nm scheme can only operate at low power, and the safe detection distance does not exceed 200m, but its cost is relatively low.
1550nm away from the visible wavelength of the human eye, most of the light will be absorbed by the transparent part of the eyeball before reaching the retina, the safety power limit is 40 times that of 905nm, and the safe detection distance can reach 250 meters, or even more than 300 meters, but it requires the use of fiber lasers, and the cost is higher than 905nm.
In general, 905nm and 1550nm lasers have their own advantages and disadvantages at the current point in time, and the two bands are a complementary coexistence state for automotive sensors.
905nm vs. 1550
Mechanical lidar still occupies a major position in the industry, and semi-solid/solid-state has good prospects. According to the classification of scanning methods, lidar is mainly divided into mechanical lidar, semi-solid-state lidar and solid-state lidar. In the long run, solid-state lidar has great potential in terms of cost and stability due to the absence of movable parts, and is the best solution in technology. Among the three current technical routes, mechanical is the most commonly used and has been widely used in Robotaxi and other fields; Hybrid lidar is a compromise between mechanical and pure solid-state (more mechanically scanning only a range within an angle ahead; The more pure solid-state still has some smaller moving parts), which is the mainstream product for mass production and loading of passenger cars at this stage.
The development of mechanical radar is relatively mature, but it is difficult to achieve vehicle-level mass production due to cost and redundant components. The technical scheme of mechanical lidar is mainly a high-line number mechanical scheme. The mechanical lidar that drives the overall rotation of the opto-mechanical structure through the motor is the classic technical architecture of lidar, and the innovation points of its technical development are reflected in the increase of the number of system channels, the expansion of the ranging range, the improvement of spatial angle resolution, and the improvement of system integration and reliability. Compared with semi-solid and solid-state lidar, mechanical rotary lidar has the advantage of 360° horizontal field of view scanning of the surrounding environment, and has stronger ranging capabilities within the field of view. However, the volume and weight of rotating parts are large, and the high-frequency rotation and complex mechanical structure make the rotating parts inside them easy to be damaged, and the service life is relatively short, which is difficult to meet the strict requirements of vehicle regulations. In addition, it achieves high wiring harnesses by increasing the number of transceiver modules, which makes the cost higher and limits its large-scale use.
Diagram of the internal structure of mechanical lidar
The semi-solid solution mainly includes micro-galvanometer (MEMS) scheme and rotating mirror scheme. The semi-solid solution is characterized by the decoupling of the transceiver unit from the scanning component, the transceiver unit (such as laser, detector) no longer carries out mechanical movement, and part of the field of view (such as forward) detection is realized by the activity of the scanning component, and the volume is more compact than that of mechanical rotary radar. The maturity of the rotary mirror solution is relatively high, and the reliability has been verified by the vehicle regulations.
The rotating mirror scheme fixes the transceiver module, and uses a 360° high-speed rotating multi-sided prismatic mirror to reflect the beam and complete the full field of view scanning within the field of view of the lidar. The advantage of the rotary mirror is that the prism, motor and transmitter have better heat resistance and durability, so it is easier to pass the car gauge, and the current Scala 1 with Valeo mirror solution has been approved by the car specification. The rotating mirror is regarded as the only way to mechanically march to pure solid-state, and it is the mainstream of short-term vehicles, and semi-solid and pure solid state will be parallel for a long time to come.
MEMS radar is limited by the galvanometer deflection range, the field of view angle is small, and the high mass production brings low cost advantages. MEMS galvanometer is a silicon-based semiconductor component, which belongs to a solid-state electronic component, which integrates a very compact micrometer on a silicon-based chip, and its core structure is a very small cantilever beam, and the deflection of the lens is achieved by the jitter of the cantilever beam. MEMS galvanometers get rid of mechanical motion devices such as motors, and millimeter-sized galvanometers greatly reduce the size of lidar. Due to its high level of integration, it is expected to have greater advantages in terms of cost and reliability as the process matures. The technological innovation of MEMS scheme is reflected in the development of larger caliber, higher frequency, and better reliability than galvanometer to apply to the technical solution of lidar. But now the MEMS deflection angle on the market is only 10-30 degrees, in order to solve the problem of small field of view, multiple transceiver modules are often spliced.
Solid-state solutions without mechanical components make it easier to pass vehicle regulations, but the technological maturity is relatively low and further development is needed. The solid-state solution is characterized by no longer containing any mechanical moving parts, and is suitable for the detection of part of the field of view (such as forward), including Optical Phased Array (OPA) scheme, Flash scheme, electronic scanning scheme, etc. Because it does not contain mechanical scanning devices, the internal structure is the most compact compared to other architectures, which has advantages in terms of size.
OPA is still in its infancy, difficult and expensive to manufacture. Optical phased array technology (OPA) adjusts the phase relationship of each phased unit by applying voltage, and uses the coherent principle to realize the deflection of the emitted beam, so as to complete the scanning measurement of a certain range of space by the system. In OPA systems, optical phase modulators are used to control the beam of light passing through the lens. OPA has the advantages of high precision, fast scanning, small size, high integration and high degree of mass production standardization, and has strong technical advantages, but because the current OPA industry chain is still in its infancy, and the manufacturing process is complex, there are still problems in mass production, and because of its complex structure, there are also problems such as high control complexity and high power consumption.
Flash lidar can quickly record scenes, but its application is limited due to the lack of detection range. Flash-type lidar is classified as solid-state lidar due to the absence of a scanning system and mechanical moving parts. Flash-type lidar can quickly record the entire scene by emitting large-coverage array lasers in all directions in a short time, avoiding the various troubles caused by target or lidar movement during scanning. It behaves more like a camera, with laser beams diffused directly in all directions, illuminating the entire scene with a single flash of light. The system then uses tiny sensor arrays to acquire laser beams that are reflected back in different directions. The disadvantage is that once the propagation distance exceeds tens of meters, the return photons are greatly reduced, making reliable detection impossible, and also increasing the height requirements for the receiving end and power, and increasing the cost.
LiDAR integration connects the upstream and downstream of the industrial chain and has strong industrial added value. LiDAR mainly includes four major systems: laser emission, scanning system, laser reception and information processing, and the different electronic components and optical systems required by the four systems together constitute the upstream of the industrial chain. Specifically, the upstream industry chain of the lidar industry mainly includes lasers, detectors, scanning mirrors, FPGA chips, analog chips, and optical component manufacturers and processors, which are the cornerstone of the laser industry and have high entry barriers; The midstream of the industrial chain uses upstream laser chips and optoelectronic devices, modules, optical components, etc. as pump sources for the manufacture and sales of various lidars; The downstream of the industry chain is mainly the application field of various lidars, including driverless cars, advanced assisted driving, service robots, surveying and mapping, high-precision maps, etc. LiDAR industry chain companies have a clear division of labor, and midstream integrated enterprises have played a role in the industrial chain and have a strong industrial position.
LiDAR industry chain
The upstream of the industrial chain is dominated by foreign manufacturers, and the gap between downstream domestic and foreign manufacturers continues to narrow. The core components upstream of lidar are lasers and detectors, and foreign suppliers have been working in the laser and detector industry for a long time and have competitive advantages in products. Domestic suppliers have developed rapidly in recent years, and domestic lasers and detectors that have passed vehicle certification have been listed. The downstream industry chain of lidar is mainly divided into industries such as unmanned driving, advanced assisted driving, service robots and Internet of Vehicles according to the application fields. Foreign unmanned technology research started early, and still has a certain leading advantage over China, but domestic unmanned technology research has developed rapidly, and there are continuous application pilots and projects landing, and the gap with foreign companies is constantly narrowing; Thanks to the high maturity of the domestic express delivery and instant delivery industry, the domestic technology development level in the field of service robots is comparable to that of foreign countries, and domestic enterprises have more advantages in terms of the richness of robot types and the diversity of landing scenarios; The Internet of Vehicles industry is more rapidly developing than foreign countries under the strong promotion of national policies such as "new infrastructure".
LiDAR industry chain
The cost of laser transceiver modules accounts for a large proportion of the cost of LiDAR, and the overall cost is expected to further decline with the advancement of mass production. According to the data of the heart of the car, the cost of Velodyne's mechanical lidar VLP-16 after disassembly of lasers, detectors, optical components, circuit boards, motor housings and structural parts accounted for 40%, 35%, 10%, 10%, and 5%, respectively. Taking Valeo Scala rotary mirror lidar as an example, the total cost of laser transceiver-related modules, laser plates, mechanical mirrors and mechanical laser components accounts for 46%. Whether it is mechanical or semi-solid-state lidar, the cost of laser transceiver related modules is high, partly because the current overall lidar shipments are small, the fixed cost is relatively high, and the overall cost of the product is expected to further decline with the advancement of lidar mass production.
At present, the field of surveying and mapping dominates downstream applications, and the field of automobile driving is expected to become the main force in the future. LiDAR has a wide range of downstream applications, mainly involving unmanned driving, advanced assisted driving, service robots, smart cities and surveying and mapping industries. According to Yole Intelligence's "2022 Automotive and Industrial LiDAR Application Report", terrain mapping in lidar applications is still the largest application field in 2021, accounting for 60% of the market share; It is followed by the industrial sector with a 27% share; Areas such as driverless taxis, ADAS (Advanced Driver Assistance Systems), wind energy, and defense accounted for the remaining 13%. However, in recent years, with the policy support for intelligent driving in various countries around the world and the rapid development of the automotive lidar industry, the penetration rate of lidar in unmanned driving and advanced assisted driving has shown a rapid growth trend. Frost & Sullivan predicts that advanced assisted driving and unmanned driving will become the main drivers of downstream applications by 2025, accounting for 34.64% and 26.30% of the lidar market, respectively, and the automotive lidar field will contribute 61% to the growth of the overall market.
Two Intelligent and electrified two-wheel drive, lidar market is broad
ADAS (Advanced Driving Assistance System) can use a variety of sensors installed on the vehicle (millimeter wave radar, lidar, mono/binocular cameras, and satellite navigation) to collect data, and combine the map data for system calculation, so as to predict the possible danger for the driver in advance and ensure driving safety. ADAS technology greatly reduces driving complexity, with features including lane monitoring, emergency braking, stability control, and more. ADAS is the first step to autonomous driving, and ADAS needs to be popularized to achieve autonomous driving.
Schematic diagram of ADAS functionality
The "three electricity" system strengthens the synergistic relationship between intelligent vehicles and electric vehicles, and the growth of electric vehicles is expected to drive the growth of intelligent vehicles. Compared with the complex mechanical system and electronic control system of fuel vehicles, the "three electric" system composed of battery, motor and electronic control of electric vehicles makes the control logic of the vehicle simpler, and has greater openness in the design architecture, which naturally matches the development of automotive intelligence more closely than fuel vehicles. With the rapid development of the electric vehicle industry, smart cars are also expected to usher in growth.
Comparison of the "three electric" systems of new energy vehicles and fuel vehicles
Driven by various factors such as subsidies and product power improvement, the penetration rate of new energy electric vehicles in mainland China has grown rapidly. Since 2022, due to the continuous strengthening of the supply side and the impact of a number of new energy promotion policies such as state subsidies and purchase tax exemption, new energy vehicle sales have ushered in rapid growth. According to the China Association of Automobile Manufacturers, 22H1 achieved sales of 2.6 million new energy vehicles, with a market penetration rate of 21.6%, completing the target of 20% new energy vehicle penetration rate in 2025 proposed by the State Council in the "New Energy Vehicle Industry Development Plan (2021-2035)" ahead of schedule. Looking ahead, with the further advancement of vehicle electrification and intelligence, the penetration rate of new energy vehicles in mainland China is expected to further increase, according to Yiou think tank research data, it is expected that the sales of new energy vehicles in mainland China will reach 11.376 million units in 2025, with a market penetration rate of more than 37%.
Mainland new energy vehicle sales and penetration rate/10,000 units
Autonomous driving can be divided into 6 levels, and the automotive industry is currently in the transition stage of L2 to L3. On August 20, 2021, China's Ministry of Industry and Information Technology issued the national standard "Classification of Automotive Driving Automation", which was determined to be implemented from March 1, 2022, which is generally consistent with the Society of Automotive Engineers (SAE) classification standard SAEJ3016. Specific to the classification level, IDC believes that L3 level is an important watershed in the degree of intelligent driving of autonomous driving, and the main decision-making responsibility of autonomous driving at L3 level and above is transferred from the driver to the operating system. In February 2020, 11 ministries and commissions, including the National Development and Reform Commission, the Cyberspace Administration of China, and the Ministry of Industry and Information Technology, jointly issued the "Intelligent Vehicle Innovation and Development Strategy", pointing out that it is necessary to ensure the large-scale production of Level 2 autonomous driving by 2025 and the market-oriented application of Level 3 in specific environments from multiple dimensions. Currently, the autonomous driving level of Continental's production vehicles is transitioning from L2 to L3.
Driving automation classification standards
Vehicle electrification drives intelligent development, and ADAS has good prospects. According to IDC data, the CAGR of China's ADAS market shipments from 2021 to 2025 can reach 15.44%; Level 1-L5 autonomous vehicle shipments are expected to reach approximately 13.62 million units by 2025, with Level 3-L5 autonomous driving expected to have an estimated 2.5% market share in 2025. With the rapid development of ADAS, the market size is also expected to usher in rapid growth, according to the data of the China Association of Automobile Manufacturers, the market size of ADAS main functions in 2020 is 84.4 billion yuan, and the market size is expected to reach 225 billion yuan by 2025.
The autonomous driving system consists of three parts, and the perception layer is the core entrance to information acquisition. Autonomous driving can be divided into three levels: perception layer, decision-making layer, and execution layer. Among them, the perception layer is responsible for the acquisition of information, which obtains the collection of surrounding information through various sensors, and transmits the information to the central processing unit (decision-making layer), which judges and makes decisions, and the execution layer performs braking, driving, steering and other operations. Sensors in the perception layer entering information accurately and error-free is the first step towards ADAS and plays a critical role in ADAS.
ADAS system configuration
The proportion of non-autonomous vehicles continues to decrease, and the penetration rate of intelligent driving at all levels is growing rapidly. According to RolandBerger, by 2025, the penetration rate of global L1+L2 intelligent driving functions will reach 76%, of which the penetration rate of L2 functions will reach 36%, and the market penetration rate of L3 and above will reach 9% in 2025, achieving a breakthrough from scratch. The share of cars without autonomous driving will drop from 42% in 2020 to 14%. In the future, as intelligent driving functions are accepted by more and more people, the market penetration rate of intelligent driving is expected to increase rapidly and achieve higher market share.
Global penetration rate of different levels of intelligent driving from 2020 to 2025
China's autonomous driving market is developing rapidly, and sensors in the perception layer contribute an important market increment. According to the report data of "Adapting to the General Trend of Change and Embracing Innovation Opportunities - Global and Chinese Automotive Industry Outlook under the Disruptive Trend" jointly published by Chebai Think Tank and Roland Berger, it is expected that China's autonomous driving market will develop rapidly in the future, and by 2030, China's autonomous driving vehicle system market size will reach about 500 billion yuan, of which sensors, chips and software algorithms are the main contributors, and the sensor market is expected to reach a market size of nearly 131.2 billion yuan by 2030. It accounts for 26.5% of the total autonomous driving market.
Autonomous driving and sensor market size/100 million yuan
The pure vision solution is parallel to the hybrid sensing solution, and the hybrid sensing solution using lidar is now being adopted by more manufacturers. Different sensors can be combined to form different sensor schemes, and there are currently two main schemes.
One is a pure vision scheme led by cameras, with millimeter-wave radar, and using vision chips to realize automatic driving; In the pure vision scheme, the camera is equivalent to the human eye, and the captured image is used by the deep learning neural network for pixel segmentation, object classification, model calibration and target tracking, which has the advantage of high resolution and can directly obtain real-world scene information recognition. However, over-reliance on data accumulation in relevant scenarios and corresponding deep learning sometimes fails at the existing technical level.
The other is a hybrid sensing scheme based on lidar, equipped with millimeter-wave radar, ultrasonic sensors and cameras. At present, the hybrid sensing scheme integrates a variety of sensor data, complements the advantages between sensors, has stronger ability to detect long-distance all-round and resist environmental influences, and has higher measurement accuracy, but the cost is relatively high. In recent years, with the breakthrough of key technologies and the decline in the cost of optoelectronic semiconductors, the volume and price of lidar are constantly decreasing, and the hybrid sensing scheme based on lidar has become the choice of more car manufacturers at this stage.
Comparison of two autonomous driving sensor solutions
LiDAR is one of the core sensors of autonomous driving, with obvious advantages in performance, anti-interference and information volume. At present, the mainstream self-driving car sensors used in the industry are vehicle cameras, ultrasonic radar, millimeter wave radar, lidar, etc. Since different sensors have different principles and functions, they can play their specific advantages in different scenarios, so it is difficult to completely replace each other for the time being. However, compared with millimeter-wave radar, lidar can achieve longer detection distance and higher accuracy; Compared with cameras, lidar detection requires less algorithm and computing power, and can distinguish between slow-moving people and other stationary objects. Compared with ultrasonic radar, lidar has a longer measurement distance and can be used normally at high speeds. With the increasing requirements for sensors for high-level autonomous driving, lidar will be an important supplement to traditional sensors to support the information acquisition of autonomous driving.
Primary sensor comparison
As ADAS continues to evolve, the demand for lidar as one of the windows to obtain environmental information has also increased. According to the data of Guanyan World, under normal circumstances, in the L3 level of the car, the number of lidar is required to be 1; In the L4/L5 level, the number of lidar is required to be 2-3 and not less than 4, respectively. Although the number of lidar mounted on vehicles in practical applications is often determined by the design of specific models, there is no obvious law in the number, but it is still positively correlated with the level of ADAS driving as a whole.
Lidar is installed on all major models
At present, 905nm and mechanical lidar are dominant, and 1550nm and semi-solid-state MEMS lidar have good prospects.
From the wavelength point of view, according to the wavelength emission of radar laser emitter, it can be divided into 340nm, 640-670nm, 750-780nm, 885nm, 905nm, 1064nm, 1550nm and other bands. Due to low cost and mature technology, lidar systems using 905nm lasers have dominated the market and become the first choice for most lidar manufacturers. However, 1550nm fiber lasers have high power, good beam quality, and higher human eye safety thresholds, and are expected to gradually become mainstream after the cost is reduced in the future with the continuous advancement of technology.
From the perspective of technical route, lidar can be classified according to the technical path of the scanning system, which can be divided into mechanical, hybrid solid-state, and all-solid-state. LiDAR manufacturers generally start mechanically and generally develop towards solid-state development. Semi-solid-state lidar is mainly divided into rotating mirror lidar, MEMS galvanometer lidar and dual prism lidar, of which MEMS lidar is currently the most landed product, and the maturity has met the mass production.
In the future, consumption upgrades and increased demand for unmanned driving are expected to drive the rapid expansion of the lidar market. LiDAR application scenarios are wide, from the perspective of market segment unmanned driving field, according to Yole data, the global LiDAR shipments in the unmanned market will increase from 140,000 to 1.3 million from 2020 to 2025, with a CAGR of 56.2%, and is expected to reach 7.4 million by 2032; In terms of sales, the global lidar sales in the driverless market were about $1.2 billion in 2020 and are expected to exceed $8.2 billion by 2032.
Global lidar sales and shipments in the field of unmanned driving from 2020 to 2032 (right axis)
The market penetration of ADAS has steadily increased, and shipments of lidar as one of the core sensors are expected to further grow. With the continuous breakthrough and upgrading of intelligent technology, the market penetration rate of ADAS will be further improved, thereby making the lidar market grow. According to Yole data, global lidar shipments in the ADAS market will grow from 200,000 in 2020 to 3.4 million in 2025, with a CAGR of 76.2%; Shipments are expected to reach 26.6 million units by 2032. Global lidar sales in the advanced driver assistance systems market were approximately $95 million in 2020 and are expected to exceed $9.6 billion by 2032.
Global LiDAR Sales and Shipments in Advanced Driver Assistance, 2020-2032 (Right Axis)
LiDAR is widely used in automotive and industrial applications, and ADAS and autonomous driving benefit from the rapid growth of downstream demand, and the proportion of downstream demand continues to increase. According to Yole, the lidar market in automotive and industrial applications is expected to grow from $1.8 billion in 2020 to $5.7 billion by 2026, with a CAGR of 21% over the period. In the automotive market, radar in ADAS accounted for 1.5% of the automotive and industrial LiDAR market in 2020, and this share is expected to be 41% in 2026, reaching a market size of $2.3 billion. The robotic vehicle segment, which includes driverless taxis and driverless shuttles, is expected to reach $575 million by 2026, with a CAGR of 33%.
Global automotive and industrial lidar segment size/million dollars
Boosted by the rapid development of various application fields, the global lidar market is expected to usher in rapid growth. With the continuous breakthrough and upgrading of intelligent technology, driven by the expansion of unmanned fleets, the increase in the penetration rate of lidar applications in advanced assisted driving, and the demand for service robots and intelligent transportation construction, the lidar market is expected to expand rapidly. According to Sullivan's data, the global lidar market reached $2 billion in 2021, a year-on-year increase of 100%, and the global lidar market is expected to reach $13.54 billion in 2025, with a CAGR of 64.6% from 2019 to 2025.
Global lidar market size and growth rate from 2017 to 2025/billion USD
LiDAR technology continues to upgrade, which is expected to drive the price of lidar down. In the early stage, due to the small overall usage of lidar and the relatively high fixed cost, the overall price is higher. Specific to each type, the early mechanical lidar is limited by its own complex structure, so the overall cost is the highest, and the price can reach tens of thousands of dollars; After the development of semi-solid-state lidar in recent years, the technology has been relatively mature, the cost has been reduced to about 1000-1500 US dollars, and the cost is expected to be further reduced with the large-scale use of car companies in the future; Solid-state lidar has the lowest cost, only about $200-500, but in a short period of time, because the technology is not yet mature, it is difficult to promote large-scale popularization. In the future, as the technology matures further, it is expected to bring down production costs, thereby driving prices down.
The logic of mutual promotion between the decline in lidar prices and the increase in shipments has been initially established, and product prices are expected to further decline. At present, there is a gradual consensus in the industry that when the scale of lidar reaches the order of 100,000, the price will drop to about $1,000. Taking Valeo as an example, from its published data, 160,000 units were shipped in 2021, and the price is already below $1,000. With the cost reduction brought about by technology upgrades and the growth of demand for lidar, the production scale of lidar is expected to continue to expand. According to the data of Guanyan World, the average unit price of lidar is expected to drop from 6500 yuan in 2021 to 4550 yuan in 2022, and it is expected that the unit price of vehicle lidar will further drop to 1719 yuan by 2030.
The increase in the number of lidar bicycles and the decline in price have led to an overall decline in the value of lidar vehicles. According to the data of Guanyan World, L3 level lidar models are equipped with about 1 lidar on average, and L4/L5 level lidar models are equipped with no less than 4 lidar on average. With the continuous improvement of technology lidar penetration rate, the average unit price of vehicle lidar is declining, the consumption of bicycle lidar is expected to increase, it is expected that by 2025, L3 level lidar models will have an average of about 2 lidars, and L4 & L5 level lidar models will be equipped with an average of 5 lidars; In 2030, L3 lidar models will have an average of about 2.5 lidars, and L4 & L5 lidar models will have an average of 5.5 lidars. Affected by the decline in unit prices, although the number of lidar mounted increased, the overall value of bicycles declined.
It is estimated that the size of China's automotive lidar market and the global automotive lidar market size will be $2.26 billion and $4.89 billion, respectively, in 2025. In 2030, the size of China's automotive lidar market and the global automotive lidar market size will be $6.94 billion and $13.57 billion, respectively. The main assumptions are as follows:
1) Passenger car sales: According to OCIA data, global passenger car sales were 56.4 million units in 2021 and 21.48 million units in China in 2021. The China Passenger Car Association expects passenger car sales in China to grow by 5% in the next few years, and at this rate, China's passenger car sales in 2025 will be 26.11 million units. Considering that after that, the demand growth space is limited and the market development momentum is insufficient, giving a 2% growth rate from 2025 to 2030, it is estimated that China's passenger car sales in 2030 will be 28.83 million units. In terms of global passenger car sales, the global automotive industry has experienced a certain decline in recent years, coupled with the impact of the new crown epidemic and war conflicts on the production and demand of the global automotive industry in 2020, global automobile production continues to be sluggish, considering the downturn in the general environment and weak demand, we estimate that global passenger car sales in 2025 and 2030 remain unchanged, and the global passenger car sales value in 2021 is expected to be 56.4 million units.
2) Penetration rate of smart vehicles in various categories: Refer to the "China Autonomous Driving Market and Future Mobility Market Outlook" report released by IHS Markit on the penetration rate of smart vehicles in various categories in China, assuming that the global penetration rate of smart vehicles is the same.
3) Bicycle lidar carrying volume and single lidar value amount: Use the above predicted bicycle radar value and vehicle carried data. The above forecast for the value of a single lidar in 2025 and 2030 is 2475 yuan and 1719 yuan, which is equivalent to about 354 US dollars and 246 US dollars at the exchange rate 7.
Automotive radar market size estimate
Third, the competition of terminal manufacturers is fierce, and various technical routes are blooming
The foreign lidar industry started early and has advantages in technology and customer base. The traditional application field of lidar is dominated by traditional surveying and mapping technology companies such as Trimble, Hexagon, Topcon and Sick, while automotive lidar as an emerging field, there are many new entrants, including Velodyne, Luminar, Aeva, Innoviz, Ouster and so on. Among them, Velodyne has significant advantages in the field of mechanical radar, Luminar and Innoviz are more advanced in MEMS lidar, Aeva first adopts the FMCW technology path, and Ouster mainly takes the Flash route.
The manufacturers with a high market share are mainly foreign manufacturers, and Sagitar Juchuang is the Chinese enterprise with the highest proportion. According to the "LiDAR 2021 for Automotive and Industrial Applications" report released by Yole, the Chinese lidar company RoboSense ranks first in China and second in the world with a market share of 10%; The largest share of the global lidar supplier market is Valeo, a traditional Tier1 manufacturer, accounting for 28%. Other companies include North American lidar companies Luminar and Cepton, veteran Tier1 companies Denso and Continental, and DJI's Livox, but they all account for less than 10%.
Market share by vendor
Automotive lidar is still in the early stage of the industry, and all major manufacturers are in the stage of rapid increase in the number of fixed points. From the perspective of fixed points, due to the high price of lidar in the early stage and the long period required for automotive-level certification, automotive lidar is still in the early stage of the industry, and the current major lidar manufacturers are still relatively few. Judging from the current fixed point situation of each manufacturer, the customers of overseas manufacturers are mainly overseas traditional car companies such as Mercedes-Benz and Audi; Domestic enterprises are mainly domestic car companies and new forces in car manufacturing, among which Sagitar Juchuang is currently the largest number of domestic fixed-point manufacturers, with certain competitive advantages.
Comparison of the current situation of major lidar manufacturers
1 , Introduction of overseas terminal manufacturers
Valeo (Valeo) :Founded in Saint-Ouen, a suburb of Paris, France, Valeo is an industrial group specializing in the design, development, production and sales of automotive parts, systems and modules. In the early days, under the authorization of the British company Ferrodo, it was mainly engaged in the sales of brake linings and wipers. The company began diversifying in 1960 and its product range continued to expand: brake systems (1961), air conditioning systems (1962), lighting systems (1970) and automotive electronics systems (1978). The company's business involves front-loading business and after-sales business, and is the world's leading auto parts supplier, providing supporting all major automobile factories in the world. Valeo consists of four divisions (Powertrain, Thermal, Comfort & Driver Assistance Systems and Vision Systems) and the Aftermarket division. In 2021, the company achieved revenue of 17.262 billion euros, an increase of 5%; Net profit attributable to owners increased by 116% to EUR 175 million.
Valeo has been working in the lidar industry for many years. The company began to develop the lidar product SCALA 1 for production vehicles in 2010, and achieved mass production in 2017, which is the earliest lidar model to pass the automotive certification. Valeo released the third generation of SCALA in November 2021, which is expected to go into production in 2024, and compared to the second generation of lidar, the third generation has a higher level of performance: 12 times more resolution, 3 times more detection range and 2.5 times longer field of view. As of the end of February 2022, Valeo has produced more than 160,000 lidars, making it the only supplier on the market to achieve mass production of automotive-grade lidars.
Velodyne:Velodyne was founded in 1983 and is headquartered in San Jose, California. At the beginning of its establishment, the company's main business was audio, and then gradually expanded its business to the field of lidar. In 2016, Velodyne spun off its core business lidar division to form a new company, Velodyne LiDAR. With the development of mechanical lidar, Velodyne Lidar has become one of the major players in the lidar industry. Through years of deep cultivation and development, the company has established cooperative relationships with global autonomous driving leaders such as Google, General Motors, Ford, Uber, and Baidu.
The company's products provide "hardware + software" integrated solutions based on customized chips. Velodyne provides smart, powerful lidar solutions for autonomous vehicles, driver assistance, delivery solutions, robotics, industrial, infrastructure, navigation, mapping, and more.
The company's products mainly include surround view hybrid solid-state lidar and directional solid-state lidar, of which the mass production of surround view hybrid solid-state lidar products include HDL-64E/32E, Puck, UltraPuck and AlphaPrime series; The product mass-produced for directional solid-state lidar is Velarray. In addition, the company also provides software solutions.
Velodyne's main products
Luminar：Luminar was founded in 2012 in Silicon Valley. In late 2020, Luminar went public on NASDAQ through a merger with shell company Gores Metropoulos. As of December 28, 2022, Luminar had a market capitalization of $1.8 billion. Luminar has partnered with Black Forest Engineering to address the high cost of indium gallium arsenic, making it known as a low-cost lidar sensor technology. Through years of development, the company has been recognized by Toyota, Audi, Volkswagen, Volvo, Mobileye and other cooperative customers. In 2021, Luminar achieved revenue of $31.94 million, an increase of 129%, and net profit of -$238 million.
Luminar's products include hardware sensors, software systems, and complete solutions to OEMs. In terms of hardware, the company's current main products are Iris and Hydra, both of which use the ToF ranging principle and 1550nm wavelength laser light source products. Hydra is a high-performance lidar for testing and development projects, focusing on the research and development of high-precision lidar in direct sunlight, rain, snow, fog and other challenging weather conditions. The company's automotive-grade product, Iris, was put into production in 2019 and is scheduled for mass production in 2022, with prices as low as $500 for L1-L2 and about $1,000 for L3-L5 products. In terms of software, the supporting software provided by the company can cover the perception and recognition at night or in harsh environments on the highway, providing additional information for autonomous driving path planning and decision-making. In terms of solutions, Luminar is committed to providing a one-stop solution, with Sentinel, the company's software product suite, the first full-stack autonomous driving solution for mass production, capable of providing every OEM with autonomous driving and active safety features on the highway.
AEVA：Founded in 2017 by two former Apple engineers, Aeva is a technology company specializing in FMCW 4D lidar. In 2021, it merged with Inter Private Acquisition Corp. through SPAC and listed on the US stock market. As of December 28, 2022, the company has a market capitalization of $270 million, ranking second among the eight pure lidar companies listed, behind Luminar.
Aeva's main product is 4D FMCW Lidar, including hardware systems and software algorithms. Aeva's 4D FMCW Lidar uses a continuous, low-power laser beam to measure the frequency change as the waveform bounces off an object. The instantaneous velocity of each point can be detected with an accuracy of centimeters per second, and the detection distance is up to 500m, and it has strong anti-interference ability. The company has released its second automotive radar product, the Aries II, the world's first 4DFMCW Lidar with camera-level resolution.
According to the company's planning, it is expected to launch a prototype in Q2 2022 and complete SOP in Q4 2023. The company's products mainly cover three major application directions: 1)
Automotive sector: including passenger cars, commercial vehicles (e.g. Robotrunk), mobility (e.g. Robotaxi); 2）
Consumer devicesConsumer Electronics (3D Projection, AR/VR), Consumer Health Market (Contactless Health Monitoring, e.g. Pulse Rate and Respiration Rate Monitoring); 3）
Industrial automation: Provide robots, industrial metrology machines, and other automation equipment with a clearer, more accurate and cost-effective way to perceive their surroundings.
2 Introduction of domestic terminal manufacturers
Sagitar Juchuang:Founded in August 2014 and headquartered in Shenzhen, Sagitar Juchuang is the world's leading lidar company. Sagitar Juchuang product technology includes: MEMS and mechanical lidar hardware, hardware fusion technology, AI perception algorithm, etc. The company is committed to providing the market with intelligent lidar systems with information understanding capabilities through the closed loop of the three core technologies of lidar hardware, AI algorithms and chips, subverting the definition of pure information collection in traditional lidar hardware, and empowering robots and vehicles with perception capabilities beyond human eyes.
The company's main products can be divided into lidar and perception solutions. Among them, lidar includes RS-LiDAR-M1 (MEMS solid-state lidar), RS-Ruby (128-line lidar), RS-Bpearl (close-range blinding lidar), RS-LiDAR-16 (16-line lidar) and RS-LiDAR-32 (32-line lidar); Perception solution products include RS-Cube (Perception Algorithm Unit), medium and low speed lidar perception scheme RS-P1, medium and high speed lidar perception scheme RS-P2, RoboTaxi lidar perception scheme RS-Fusion-P3 and RoboTaxi lidar perception scheme RS-Fusion-P5.
The company's main customers include SAIC, Geely, FAW, AutoX, Pony.ai, etc. In 2021, Sagitar released the SOP version of the automotive-grade solid-state lidar RS-LiDAR-M1. The company's second-generation intelligent solid-state lidar RS-LiDAR-M1 has been designated by a large number of car companies, including L3 heavy-duty truck solution technology enterprises, North American new energy vehicle companies, new Chinese automakers, traditional OEMs, top supercar brands, etc., covering a variety of models from supercars to family cars, from passenger cars to commercial vehicles. Among them, GAC AION officially announced in July 2021 that it will be equipped with the M1 on a number of models. In December 2020, the M1 prototype was shipped to a North American automaker, becoming the world's first automotive-grade MEMS solid-state lidar to be delivered in bulk.
Hesai Technology:Hesai Technology was established in Shanghai in 2014. The company launched the 40-line mechanical lidar Pandar 40 in April 2017, and then released Pandora, an autonomous driving developer kit based on Baidu's Apollo platform, in late 2017, which loaded the lidar Pandar 40 and multiple panoramic cameras into the same device for synchronous perception work, realizing the underlying synchronization and fusion of lidar and other sensor data, indicating that the company can help downstream customers solve the industry pain points of tuning perception system hardware. The company is committed to becoming a provider of system solutions. After years of deep cultivation, Hesai has deep accumulation in core components, self-developed chips, automotive-grade production capacity, functional safety, active anti-interference technology and deep learning-based lidar perception.
Hesai's customers include the world's mainstream autonomous driving companies, top automakers, Tier 1 suppliers, robotics companies, etc., in 40 countries and more than 90 cities around the world. The company's current major customers include two of the top three automakers in North America, one of the top four in Germany, more than half of the top 15 DMV road test miles in California, USA, and most of China's leading autonomous driving companies.
The company's lidar products are mainly mechanical and semi-solid-state lidar based on ToF ranging method. The company's mechanical lidar products have been widely used in the field of unmanned driving, while semi-solid-state lidar is mainly for the emerging mass-produced passenger car advanced assisted driving market. At the end of 2017, the company began to deploy chip self-research, design chips according to product requirements, and the current chip-based V1.0 results multi-channel laser driver chip and multi-channel analog front-end chip have completed mass production, and are applied to a number of lidar research and development projects and Pandar, XT mass production projects. In addition, the company has also carried out the technical layout and accumulation of solid-state lidar for FMCW lidar and electronic scanning solutions, and has the product development capabilities of new technology lidar.
The company's core technology is lidar system + chip research. Based on LiDAR special chips, optoelectronic devices and micro-optical processes, the company develops mechanical LiDAR, semi-solid-state LiDAR based on micro-galvanometer scheme and rotating mirror scheme, and solid-state LiDAR based on electronic scanning scheme to meet the differentiated needs of LiDAR in different application scenarios. At the same time, the company actively develops FMCW lidar technology to prepare for the future launch of this program product.
Tudatong:Founded in 2016, Innovusion is the world's leading provider of image-level lidar. The company has R&D centers in Silicon Valley, Suzhou and Shanghai, and highly industrialized automotive-grade lidar manufacturing bases in Ningbo and Wuhan. Tudatong is committed to the exploration and innovation of lidar, and constantly provides high-performance lidar products and diversified application solutions. Its image-level ultra-long-range lidar Falcon was delivered in March 2022 as standard on NIO's ET7 autonomous driving ultra-sensing system, the world's first mass-produced 1550nm lidar.
Tudatong lidar products are divided into cheetah, jaguar, falcon series, cheetah and jaguar series products are 300-line 1550nm lidar, detection distance up to 280 meters, used in smart cities and highways, as well as rail and mining fields; The Falcon series (1550nm) is an automotive lidar that will be installed in the NIO ET7 with a detection range of up to 500 meters, which is the farthest detection distance in the current automotive lidar, with an ultra-wide viewing angle of 120° and an equivalent resolution of 300 lines.
From the customer's point of view, the lidar provided by the company for NIO has an ultra-wide viewing angle of 120 degrees, ultra-high resolution equivalent to 300 lines, an ultra-long detection distance of up to 500 meters, and a 1550mn laser to improve performance while taking into account safety indicators. In the future, the company, Joyson Electronics and NIO will cooperate in-depth in LiDAR perception fusion, V2X data fusion, autonomous driving domain controller decision-making algorithms, etc., to jointly promote the industrialization and internationalization of intelligent networked vehicles.
Radium God Intelligence:LeiShen Intelligent was established in 2015 and is headquartered in Shenzhen. The company is the world's leading lidar and overall solution provider, committed to empowering industrial upgrading with stable and reliable lidar environmental perception technology, and downstream applications covering nine industrial ecosystems such as autonomous driving, intelligent transportation, rail transit, robotics, logistics, security, surveying and mapping, ports and industrial automation. The company's main customers include Foxconn, Gree, Midea, Haier, Dongfeng Motor, Shaanxi Automobile Heavy Truck, etc. At the technical level, Leishen Intelligent is the only "Guangdong LiDAR Engineering Technology Research Center" recognized by the Department of Science and Technology of Guangdong Province, with complete independent property rights and innovative technology in lidar core hardware, special chips, and AI algorithms, it has created the most complete lidar product matrix on the market, which is not only the only lidar company in the world that has mastered the four measurement principles of ToF time flight, phase method, trigonometric method and frequency modulation continuous wave, but also the only domestic independent research and development of LiDAR 16 Channel TIA chips, lidar automated and semi-automated production lines, lidar company for 1550nm fiber lasers.
The company's products include lidar (including ToF 32-line, 16-line and single-line, MEMS solid-state lidar, long-distance lidar, laser three-dimensional imaging lidar, 3D FLASH, phased array, trigonometry, phased lidar), three-dimensional laser scanner, laser displacement sensor, high-precision laser sensor, laser mosquito killer, laser mosquito killer robot, special fiber laser and other hardware products. As well as system solutions in autonomous driving, intelligent transportation, intelligent logistics, rail transit, surveying and mapping, high-end security, bridge collision avoidance, industrial automation and other fields.
DJI:LivoX, a subsidiary of DJI, was founded in 2016. The company is committed to providing high-performance, low-cost lidar sensors. By reducing barriers to entry and production costs, Livox integrates lidar technology into more products and applications, bringing innovative changes to industries such as autonomous driving, smart cities, surveying and mapping, and mobile robots. Livox products have been sold in 26 countries and regions, including the United States, Canada, China, Japan and the European Union. Livox's major customers include autonomous driving companies AutoX and Cidi, commercial mobile robotics company Refraction AI and Gaoxian Robotics, and digital greenland focusing on air, ground and mobile laser scanning technology.
At the heart of the company's technology is non-repetitive scanning, which means that as the laser beam scans non-repetitive within the field of view (FOV), the area area scanned by lidar increases over time. Compared with traditional mechanical lidar linear repetitive scanning, the non-repetitive scanning method has three main advantages: 1) the scanning trajectory will not be repeated; 2) It can achieve nearly 100% field of view coverage with the increase of scanning time; 3) There is no rotational wear of electronic components, and the reliability is higher. Overall, non-repetitive scanning reduces the probability of objects in the field of view being missed, helping to detect more detail in the field of view. In this way, Livox not only ensures product reliability, but also reduces the cost of lidar to a certain extent.
Huawei:The development of Huawei's lidar products began in 2016. As an important part of the four ecosystems built by Huawei's MDC intelligent driving platform, lidar is also an area that Huawei is laying out. Huawei's team visited a large number of TOP car companies and clarified its product research and development direction, to make a high-performance, automotive-grade lidar that can be mass-produced. Based on this positioning, the lidar team launched the "climbing the north slope strategy", and the company did not choose to start from the traditional rotating mechanical lidar, but chose to develop from the difficult front-loaded mass production products. According to its official information, Huawei has been advancing according to the annual output of 100,000 sets / line to meet the needs of large-scale mass production in the future.
Huawei's first automotive-grade 96-line mid- and long-range lidar lidar has a large field of view of 120° horizontally and 25° longitudinally, which can meet the needs of people and vehicles in urban areas and high-speed scenarios, and the full-field ranging of the product can reach 150 meters. Compared with similar products, Huawei's lidar is at the industry-leading level in FOV and angular accuracy.
Chedong believes that although the foreign lidar market developed earlier, the current domestic lidar market has begun to show a hundred flowers, and many start-ups are more advanced in mechanical radar and are developing towards solid-state radar. In the future, with the reduction of the cost of mass production LiDAR, for safety reasons, LiDAR is expected to become the standard for mid-to-high-end automobiles, and the scale of the automotive LiDAR market is expected to further expand.