In-vehicle sensor technology and industry chain review, autonomous driving is accelerating penetration

Source: China Galaxy Securities Research Institute

If you want to get the PDF document, click the blue word " Smart Car Expert " above and reply "Sensor 2" to receive it

1. The autonomous driving market is vast, and domestic and foreign manufacturers are accelerating their layout

The global autonomous driving market is vast. With the gradual popularization of artificial intelligence and 5G technology, unmanned driving and advanced driver assistance are developing rapidly. The realization of these technologies can greatly reduce the traffic risks caused by human errors, improve transportation efficiency, enhance road capacity, change the production and consumption patterns of automobiles, and realize the vision of safe, efficient and green development of transportation.

According to IHS's forecast, self-driving cars will begin a round of explosive growth around 2025. By 2035, half of the vehicles on the road will be self-driving. By then, the total revenue scale of self-driving vehicles and related equipment and applications will exceed US$500 billion.

At present, most autonomous driving manufacturers can basically achieve L2 autonomous driving and have begun to launch models with preliminary L3 functions. Among foreign automakers, the most technologically advanced autonomous driving car is Google's Waymo, which is equipped with multiple laser radars, millimeter-wave radars, cameras and high-precision automatic positioning devices, and the whole vehicle can achieve L5 autonomous driving.

2. Sensors are the cornerstone of autonomous driving and will see rapid growth

According to the different working mechanisms and purposes of automotive sensors, sensors can be divided into traditional sensors and intelligent sensors. Traditional sensors act as automotive neurons to control various systems of the car. Common types include: pressure sensors, position sensors, temperature sensors, etc.

These traditional sensors sense the specified physical quantity and convert it into a usable input signal according to certain rules, converting non-electrical quantity into electrical quantity. The information collected by them is processed by the electronic control unit to form execution instructions and complete electronic control.

Smart sensors are the core of autonomous driving. Currently, sensors used for autonomous driving environment perception mainly include: millimeter wave radar, laser radar, ultrasonic radar and vehicle-mounted cameras. Cameras are the basis of traditional visual solutions, with low prices and can be installed in different locations according to different functional requirements.

Millimeter-wave radar refers to radar that works in the millimeter-wave band, with a wavelength of 1 to 10 mm, between microwaves and centimeter waves, and has the advantages of both microwave guidance and photoelectric guidance. Ultrasonic radar uses the time difference from ultrasonic emission to reflection reception to calculate the distance to obstacles, and is often used in parking systems.

LiDAR mainly detects the position, speed and other characteristic quantities of the target by emitting laser beams. According to the State Intellectual Property Office, the top three sensors in the world in terms of the number of patents for autonomous driving sensors are visual sensors, millimeter-wave radars and LiDARs, accounting for 30%, 22% and 20% respectively.

Multi-sensor information fusion is the only way to achieve autonomous driving. Different types of sensors have obvious advantages and disadvantages. A single sensor cannot meet the complex application scenarios of autonomous driving. Multi-sensor information fusion has become an industry consensus. Multi-sensor information fusion (MSF) uses computer technology to perform multi-level and multi-space combined processing of information and data from multiple sensors or multiple sources, and ultimately makes judgments and decisions.

In this process, different sensors complement each other and play their respective functions in different usage scenarios, effectively improving the redundancy and fault tolerance of the system, and enhancing the accuracy and intelligence of system decision-making. According to different information processing methods, the architecture of multi-sensor information fusion can be divided into centralized, distributed and hybrid. The hybrid combines the advantages of centralized and distributed, and is widely used in practical occasions.

Judging from the autonomous driving perception solutions of representative models of mainstream car companies, a variety of sensor fusion solutions are widely adopted. Take the General Motors Cruise AV as an example. The general goal is to achieve L4 level autonomous driving. The whole car is equipped with 5 Velodyne VLP16 16-line laser radars, 21 millimeter-wave radars (12 of which are 79GHz millimeter-wave radars provided by Japan's ALPS) and 16 cameras.

However, domestic autonomous driving car manufacturers currently mostly use a combination of cameras, millimeter-wave radars and ultrasonic radars. Due to the high cost of lidar, low domestic market penetration and limited application scenarios, it is currently less used in China.

The accelerated penetration of autonomous driving will drive the rapid growth of the sensor market. The development of autonomous driving has created huge development opportunities for the core component industry represented by vehicle-mounted cameras, millimeter-wave radars and lidars. Among them, ADAS, as an important application field of automotive sensors, has a direct role in promoting the development of the sensor market through its scale expansion.

The rapidly expanding autonomous driving market has led to an increase in demand for automotive sensors, which has led to the continuous expansion of the sensor market. According to TouBao Research Institute, the market size of China's automotive sensor industry was close to 20 billion yuan in 2019, and with the promotion and upgrading of autonomous driving, the sensor market will further expand. It is expected that by 2023, the scale of China's automotive sensor market will exceed 55 billion yuan.

1. Relatively mature, with higher value than mobile phone cameras

Cameras are relatively mature and play a leading role in autonomous driving below L3. On-board cameras are an important way for autonomous vehicles to collect information and analyze images, thereby realizing a series of functions such as road sign recognition, pedestrian recognition, vehicle recognition, and lane line sensing, playing an important role in autonomous driving systems.

Compared with LiDAR, camera-based solutions are relatively mature and play a leading role in autonomous driving below L3. According to the different functional requirements and installation locations of ADAS, on-board cameras include front-view, surround-view, rear-view, side-view and built-in cameras. Cameras in different locations have different functions and are an indispensable component for realizing autonomous driving.

According to the number of cameras, they are divided into monocular, binocular and multi-camera. In the short term, monocular cameras are still the mainstream. Both single and binocular lenses obtain distance information through image data collected by the camera, and play an important role in the position of the front-view camera; but there are differences in the distance measurement principles of the two. Monocular vision calculates the distance according to the size of the target after image matching, while binocular vision measures the distance by calculating the parallax of the two images of the two cameras.

The difference in principle makes binocular cameras more accurate and precise than monocular cameras, but the cost is also higher, and they are mostly installed on high-end models. Compared with monocular and binocular cameras, multi-cameras use multiple different cameras to cooperate with each other to cover different ranges of scenes, and can more accurately identify and analyze the environment. The corresponding hardware cost and technical requirements are also higher. Currently, they are only used by some manufacturers in individual models.

Monocular cameras were developed earlier, and the technology is now relatively mature, with low mass production and promotion costs; however, due to the limitations of a single camera's fixed focus, it is difficult to switch the focus at different distances, and it is difficult to take into account the distance and range of measurement. Binocular and multi-cameras have overcome the limitations of a single camera to a certain extent, and can obtain a wider coverage range and more accurate data based on the cooperation of multiple cameras.

However, the hardware cost and corresponding algorithm requirements of multiple cameras are high, and the corresponding supporting facilities are not yet fully developed, so it is difficult to mass produce and promote them at this stage. Monocular cameras have low chip computing power and low cost, and can meet the needs below L3 when paired with millimeter-wave radar and ultrasonic radar. Therefore, monocular cameras will remain the mainstream solution for vehicle cameras in the short term.

The environmental conditions of vehicle-mounted cameras are more stringent, and they have correspondingly higher value. Compared with traditional industrial cameras and mobile phone cameras, vehicle-mounted cameras need to maintain a stable working state under various complex working conditions such as high and low temperatures, humidity, strong dim light and vibration. Therefore, vehicle-mounted cameras have higher safety level requirements and process performance requirements. Considering safety factors, automobile manufacturers tend to choose parts manufacturers with mature technology and guaranteed quality. Therefore, it takes a longer certification cycle for vehicle-mounted parts manufacturers to enter the market system and obtain ratings, and the industry barriers are relatively high.

Specifically, in order to cope with complex and diverse usage environments, vehicle-mounted cameras need to have high dynamics, and their temperature range is generally -40 degrees to 80 degrees. At the same time, they need to have high anti-magnetic and anti-vibration performance, and they must also meet the basic service life requirements of 8-10 years. In terms of performance, considering the chip processing burden, the pixel requirements of vehicle-mounted cameras are relatively low, generally 300,000-1.2 million pixels, and the power is generally below 10W.

In order to identify objects within a larger range, vehicle-mounted cameras have higher requirements for detection angle and range. Surround view and rear view generally use wide-angle lenses of more than 135 degrees, and the front view camera lens range is 40-70 degrees. Therefore, vehicle-mounted cameras have higher technical barriers and correspondingly higher value.

(II) The industry is experiencing rapid expansion, with the market size reaching US$27 billion in the next five years

The coverage rate of in-vehicle cameras is low, and the market potential is huge. To fully realize autonomous driving, the car must be equipped with five types of cameras, and the number of cameras per car must be at least 6. According to Yole data, the average number of cameras per car in the world will increase from 1.7 in 2018 to 3 in 2023, and this number will increase further with the upgrade of autonomous driving.

As vehicle-mounted camera technology matures, the price of vehicle-mounted cameras continues to decline. According to ICVTank data, the price of vehicle-mounted cameras is expected to be 145 yuan in 2020. This price is expected to drop further in the future, and will further promote the coverage rate of vehicle-mounted cameras and the number of single-vehicle configurations.

The number of vehicle configurations has increased, and the demand for vehicle cameras has continued to grow. The increase in the number of vehicle cameras configured per vehicle has strongly boosted the market demand for vehicle cameras. According to data from the China Business Industry Research Institute, global vehicle camera shipments increased from 70 million in 2016 to 120 million in 2019, and global vehicle camera shipments are expected to reach 140 million in 2021.

We expect that China's automotive camera shipments will exceed 50 million in 2021 and will grow to approximately 190 million by 2025. The market has huge potential for development.

The scale of the in-vehicle camera industry continues to expand, and the market has broad room for development. With the development and popularization of autonomous driving, the in-vehicle camera market is expected to achieve rapid expansion. According to ICVTank data, since 2015, both the global and Chinese in-vehicle camera industries have achieved significant growth, and this growth trend will continue in the future.

It is estimated that by 2025, the global automotive camera industry will reach US$27 billion, and the Chinese automotive camera industry market will reach RMB 23 billion. In the future, with the popularization of autonomous driving and the maturity of technology, the automotive camera industry will usher in an era of rapid expansion.

(III) Foreign leading enterprises have significant advantages, and domestic enterprises are rising rapidly

The industry chain of automotive cameras includes core components, module packaging and system integration, software algorithms and solutions. The core components mainly include lens groups, CMOS image sensors, optical lenses, filters, sonic motors and digital signal processing chips (DSP).

Module and system integration is the integration of various components into camera hardware. In the entire vehicle camera hardware, CMOS image sensor is the core component, accounting for 50% of the cost, while module packaging and optical lens costs account for 25% and 14% respectively.

The software of the vehicle camera mainly relies on its internal chip and the chip-based visual system autonomous driving algorithm. The algorithm of the vehicle camera mainly relies on computer vision and deep learning. The specific steps of computer vision include image input, preprocessing, feature extraction, feature classification, matching and complete recognition. Deep learning simulates the neural network of human thinking, greatly simplifies the perception process, and can achieve the leap from input picture to output result.

Considering the limitations of deep learning in retrospecting the causes, it is necessary to add a rational decision-making part to deep learning and design it in blocks. Due to the lack of vehicle-side platforms, deep learning has not yet been commercialized, but with the further development of algorithm models and algorithm types, deep learning is expected to be further popularized in the foreseeable future.

In the automotive camera lens group market, China's Sunny Optical has a clear leading advantage. Sunny Optical's automotive camera lens shipments rank first in the world. According to data compiled by the Forward-looking Industry Research Institute, Sunny Optical's market share reached 34% in 2019, followed by South Korea's Sekonix, Kantatsu and Japan's Fujifilm. The top four companies in the industry have a market share of 78% CR4. As a world-leading domestic lens manufacturer, Sunny Optical entered the automotive lens field in 2004. In 2012, its shipments ranked first in the world, becoming an industry leader.

At present, Sunny Optical's automotive products include front-view, rear-view, surround-view, side-view and interior-view lenses, and it has put forward good product solutions in the field of automotive lenses. Its main customers include Mercedes-Benz, BMW, Audi, Toyota and many other European, American, Japanese, Korean and domestic automobile manufacturers. In addition, Lianchuang Electronics, as one of the leading companies in domestic optical lenses, entered the field of automotive lenses in 2015 and achieved breakthrough development.

At present, the company has 2 lenses that have passed the MobileyeEyeQ4 certification and 8 lenses that have passed the MobileyeEyeQ5 certification. The products have been recognized by core Tier manufacturers such as Valeo, Magna, and Aptiv, and provide related products to well-known car companies such as Mercedes-Benz, BMW, and Tesla.

In the field of automotive CMOS sensors, Will Semiconductor is the world's second largest market share leader. ON Semiconductor has been deeply involved in the field of automotive electronics and ranks first in the world with a market share of more than 60% in the field of automotive CIS, which has a great competitive advantage. OmniVision Technologies and Sony rank second and third in the world, with market shares of 20% and 8% respectively. In recent years, OmniVision Technologies has continuously achieved technological breakthroughs in the field of CIS and its market competitiveness has been continuously improved.

OmniVision and Sony combined mobile phone CIS with automotive CIS and mastered large and small pixel exposure technology. In addition, OmniVision has repeatedly launched high-pixel products OX08A and OX08B, achieving a breakthrough in high-pixel in the automotive CIS field. In 2019, Will Semiconductor completed the acquisition of OmniVision. With OmniVision's advantages in technology and market share, Will Semiconductor is expected to achieve further market expansion and development in the future.

Domestic enterprises are flourishing in many places, and the trend of domestic substitution of automotive CMOS sensors is obvious. In addition to Will Semiconductor, BYD Semiconductor has achieved a breakthrough in automotive-grade CIS after years of development, launched the first 1.3 million pixel automotive-grade image sensor in China and achieved mass installation in 2018. SmartSens acquired Shenzhen Allchip Microelectronics Co., Ltd. (Allchip) in 2020 to actively expand its product line in the automotive field.

Anxin Micro has a variety of self-developed SOC series image sensor products, which are highly competitive in the field of automotive CIS. In the future, SmartSens is expected to achieve breakthrough development in the field of automotive CIS. GalaxyCore is actively deploying in the field of automotive CIS. Its products have been applied to terminal applications such as driving recorders, in-car cameras, 360-degree surround view, rear view and driver fatigue detection. It has cooperative relations with mainstream brands such as Novatek, MStar Semiconductor, JL, Fudan, and Lingyang.

The field of automotive camera modules is highly concentrated, with overseas companies occupying the main market share. Domestic companies such as OFILM and Q Technology are actively deploying and are expected to occupy a larger share in the future with their cost advantages. According to public data from China Industry Information Network, among the global automotive module packaging manufacturers, Panasonic, Valeo, Fujitsu and Continental account for 20%, 11%, 10% and 9% respectively, and the industry is highly concentrated. According to Gasgoo Automotive Research Institute, in 2020, my country's automotive camera module shipments reached 44 million, with a CAGR of 20.6%, and there is huge potential for future growth. Domestic companies are accelerating their development, and are expected to occupy a larger share in the module packaging end in the future with their cost advantages.

Mobileye is far ahead in the field of automotive camera software algorithms, and domestic companies are gradually starting to gain ground. In the field of automotive cameras, Mobileye's algorithms are mature, with a market share of more than 70%. It currently provides support to many domestic and foreign automakers such as Volvo, Audi, BMW, and Nissan. At present, my country's automotive camera algorithm companies are still in their infancy. SmartEye uses "FPGA+ARM" chips based on binocular cameras and is equipped with SmartEye's MPV algorithm. In the future, the company plans to continue to explore the field of binocular algorithms and gradually narrow the gap with leading companies.

In terms of system integration, traditional Tier 1 suppliers dominate, and products from domestic companies such as Desay SV and HUAYU Automotive have been used in domestic brand cars. Valeo, Continental, Hella, Panasonic, and Sony dominate the camera front-end market. Domestic companies have accelerated their layout in recent years. As one of the earliest companies in China to deploy in-vehicle cameras, Desay SV has achieved mass production of high-definition in-vehicle cameras and surround view systems, and currently provides in-vehicle cameras and other related supporting products to domestic brands such as Geely, GAC, and Chery;

In addition, the company has also developed a series of camera-related driving assistance systems such as automatic parking systems and night vision systems, which have been applied in models such as Geely Xingyue and Chery Jietu. As one of the largest auto parts companies in China, HUAYU Automotive has actively deployed in the field of vehicle-mounted cameras, and has realized multiple autonomous driving functions such as adaptive cruise control (ACC), forward collision warning (FCW), and blind spot detection (BSD) through visual sensing systems and external information. The company has also deployed advanced driving assistance systems such as automatic parking systems and 360-degree vehicle driving environment scanning systems, which can meet the needs of L3 and below autonomous driving.

Since its establishment, Huayang Group has been focusing on the field of automotive electronics. After years of development, it has developed into one of China's large-scale automotive electrical equipment companies with a rich product line. The company has actively deployed in the field of vehicle-mounted cameras and introduced high-definition camera automatic production lines. It has successively launched 360-degree surround view systems and "Yuyan" technology, and began to apply them to self-driving cars. Among them, camera products equipped with the "Yuyan" system entered mass production on the new Baojun E300 in 2020.

The development of vehicle-mounted cameras is relatively mature and is the most widely used "eye of the car". We believe that vehicle-mounted cameras are not only the core sensors of current autonomous driving, but will also play a leading role in future high-level autonomous driving. As the level of autonomous driving increases, the number of cameras per vehicle will continue to increase, from 1-2 in L2 to 3-6 in L3. In the future, high-level autonomous driving is expected to reach more than 10, which will drive the rapid growth of the vehicle-mounted camera market. In the next five years, the global CAGR will be about 16%, and the Chinese CAGR will be about 32%.

1. Millimeter-wave radar is on standby 24/7, and 77 GHz is the future direction

Millimeter wave radar is one of the core sensors of ADAS. Millimeter wave radar transmits millimeter waves with a wavelength of 1-10nm and a frequency of 30GHZ-300GHZ through the antenna, receives and processes the reflected signal, and then quickly and accurately obtains the physical environment information around the car body, and then performs intelligent processing and decision-making based on the detected object information. It was first used in the military field and then entered the automotive field.

The advantages of millimeter wave radar are mainly as follows:

(1) Millimeter waves have the ability to penetrate smoke, dust, and fog and can work around the clock. Compared with other sensors such as lidar, the characteristics of millimeter waves allow millimeter wave radars to be used in a wider range of environments and can be on standby around the clock.

(2) Millimeter-wave radar antennas have small apertures, high spatial resolution, are easy to carry, and have good directionality.

(3) Millimeter-wave radar has a narrow beam and good resistance to electronic interference.

(4) Low price. The current price is in the thousands of yuan range, which is lower than other sensors such as LiDARs that cost tens of thousands of yuan.

FMCW is the mainstream modulation method. According to the duration, millimeter radar waves can be divided into pulse waves and continuous waves, and the latter can be further divided into FSK (frequency shift keying), PSK (phase shift keying), CW (constant frequency continuous wave), FMCW (frequency modulated continuous wave) and other methods. At present, FMCW has become the mainstream modulation method of millimeter wave radar and is adopted by mainstream suppliers such as Delphi, Denso, Bosch, and Continental.

FMCW frequency modulated continuous wave has the advantages of being able to measure the distance and speed information of multiple targets at the same time, continuously track targets, high system sensitivity, low false alarm rate, not easily interfered by external electromagnetic noise, long measurement distance and high resolution, low required transmission power, low cost, and signal processing difficulty and real-time performance that can meet system requirements.

According to the effective range and frequency band of millimeter-wave radar, vehicle-mounted millimeter-wave radar can be divided into long-range radar (LRR), medium-range radar (MRR) and short-range radar (SRR). The frequency bands currently allocated by various countries for vehicle-mounted millimeter-wave radar are mainly concentrated in 24GHz and 77-79GHz, which belong to the range of SRR/MRR and LRR respectively.

77GHz is the future development trend. The millimeter wave radars on the market mainly include 24GHz and 77-79GHz. 24GHz is mainly used for short- and medium-range measurement functions such as blind spot monitoring and lane change assistance, while the latter is a long-range millimeter wave radar, mainly used for long-range measurement functions such as autonomous driving and forward collision warning. At present, China, Japan and other regions do not allow the use of the 79GHz frequency band, so 77GHz is the main long-range millimeter wave radar.

Compared with 24GHz radar, 77GHz is smaller in size and has stronger penetration. Its detection distance can reach 150-250 meters. Its detection accuracy is 3-5 times that of 24GHz radar. The acceptable upper speed limit is 250km/h. It has obvious advantages in recognition accuracy, measurement distance and environmental adaptability.

Therefore, the technical requirements and production costs of 77GHz are also higher, with higher technical barriers. The mainstream products on the market are currently 24GHz radars. In the future, as the technology matures and the cost decreases, 77GHz radars will gradually replace 24GHz radars and become the mainstream development trend of the future millimeter-wave radar market.

(II) Product upgrades are accelerating, and the market potential is huge

With the gradual upgrade of ADAS systems, the requirements for millimeter-wave radar will be higher. The general trend of millimeter-wave radar technology is to develop in the direction of lower cost, smaller size, lower power consumption and higher integration.

In radar chip technology, highly integrated monolithic microwave integrated circuits (MMIC) have become the mainstream; and in chip technology, the trend is towards using CMOS technology to make MMIC smaller.

In order to achieve all-round sensing coverage of the car's surrounding environment, a car needs multiple millimeter-wave radars. For example, L3 level autonomous driving requires at least 5 millimeter-wave radars (1 long and 4 short); as the level of autonomous driving increases, the number of millimeter-wave radars also increases. By the L5 advanced autonomous driving stage, the number of millimeter-wave radars will increase to more than 10.

The combination of "short-range + medium-range + long-range" millimeter-wave radars completes multiple ADAS functions such as adaptive cruise control (ACC), automatic emergency braking (AEB), forward/rear collision warning (FCW/RCW), lane change assist (LCA), blind spot detection (BSD), reversing assist (BPA), and parking assist (PA).

Driven by autonomous driving, the automotive millimeter-wave radar market has a broad space. With the popularization of autonomous driving cars in the intelligent era, autonomous driving related accessories will usher in an important growth moment. According to DIGITIMES Research data, the global millimeter-wave radar market has exceeded US$2 billion since 2017 and has maintained a high-speed growth trend. It is expected that by 2022, the global millimeter-wave radar market will reach US$16 billion.

Among them, short- and medium-range millimeter-wave radar will increase from US$1.185 billion in 2017 to US$8.4 billion in 2022, and long-range millimeter-wave radar will achieve a leap from US$1.616 billion to US$7.56 billion during the same period.

According to Yole data, China's millimeter-wave radar market grew from 1.9 billion yuan in 2015 to 7.2 billion yuan in 2020, with a growth rate of more than 30%. According to OFweek's forecast, by 2025, China's millimeter-wave radar market will exceed 30 billion yuan.

(III) Competition landscape: overseas-led, with broad room for domestic substitution

From the perspective of the global market, overseas companies still occupy a large market share and the industry concentration is high. In the global millimeter-wave radar market, the top five market players are: Bosch, Continental, Hella, Fujitsu, and Denso, which together accounted for 68% of the global market share in 2018. Among them, Bosch, Continental, Hella, and Denso are all traditional Tier 1 suppliers. Fujitsu is a world-leading information and communication technology company and also has strong strength in the field of millimeter-wave radar.

Overseas companies have different layouts in millimeter-wave radar. Looking at the main product parameters of different companies, Bosch's main millimeter-wave radar products are concentrated in 76-77GHz, focusing on medium and long-range detection, and the maximum detection distance of LLR products can reach 250 meters.

Continental is leading in 77GHz and 24GHz products, focusing on 77GHz products. The fifth-generation LRR has a maximum detection distance of 300 meters. Delphi is also one of the international companies that monopolizes 77GHz technology; while Hella mainly focuses on 24GHz frequency millimeter-wave radar and leads in short-range millimeter-wave radar technology.

Domestic enterprises are catching up quickly, with a breakthrough in 24GHz and 77GHz in the initial stage. Faced with a market that is almost monopolized by foreign-funded enterprises, domestic enterprises have stepped up their efforts in millimeter-wave radar research and development in recent years to find market breakthroughs and accelerate their pursuit of foreign capital. With the help of heavy investment in research and development, some domestic enterprises have achieved certain research and development results, and market-oriented products are about to be launched;

However, due to technical and resource limitations, the mainstream direction is still focused on 24GHz radar. In terms of 77GHz millimeter-wave radar, the core technology is still controlled by foreign component manufacturers, and the domestic industry is still in its infancy, with only a very small number of companies able to produce prototype products of 77GHz radar.

The upstream hardware of millimeter-wave radar mainly includes antenna PCB board, front-end transceiver RF components (MMIC chip), digital signal processing and control circuit, among which antenna PCB board and MMIC chip are core components.

Antenna PCB boards are the core components of millimeter-wave radars, and domestic substitution has made great breakthroughs. The mainstream solution for millimeter-wave radar antennas is currently microstrip arrays. High-frequency antenna PCB boards require high signal strength in a small integrated space. Therefore, the industry's technical barriers are high, and the cost of millimeter-wave radars accounts for about 10%. Germany's Schweizer has a global market share of 30%, leading the industry.

In recent years, domestic antenna PCB board companies have made great breakthroughs. The main manufacturers include Shanghai Electric Co., Ltd., Shenzhen South Circuit, Jingwang Electronics, Shengyi Electronics, etc. Among them, Shanghai Electric Co., Ltd. is the PCB board supplier of Continental and Bosch. Shenzhen South Circuit also has the ability to supply automotive high-frequency PCB boards. Shengyi Electronics' PCB boards for millimeter-wave radars have mass production capabilities, and potential customers include Bosch, Continental, etc., and mass production and shipment are expected to be achieved from 2021 to 2023.

The front-end transceiver RF components mainly use MMIC chip technology, which is currently dominated by overseas giants. The current mainstream solution for RF components is monolithic microwave integrated circuits (MMIC), which account for about 25% of the cost of millimeter-wave radar. MMIC chips have the advantages of low loss, low noise, wide bandwidth, high power, and strong radiation resistance. They can greatly simplify the structure of millimeter-wave radars, reduce the production cost of millimeter-wave radars, and promote the widespread use of millimeter-wave radars.

At present, the international MMIC market is mainly monopolized by foreign companies such as NXP, Infineon, and Texas Instruments (TI). Domestic chip design companies are actively deploying in the field of millimeter-wave radar, striving to narrow the gap with industry-leading companies. Domestic companies represented by I-Star Semiconductor and Calterah Microelectronics have completed the research and development of MMIC chips and are gradually entering the industrialization stage.

We believe that millimeter wave radar is one of the core sensors for autonomous driving. It has the advantage of working around the clock and will play an important role in autonomous driving. In terms of market size, it is expected that the global millimeter wave radar market will reach US$16 billion by 2022. As the technology matures and the cost decreases, 77GHz radar will gradually replace 24GHz radar and become the mainstream of the future millimeter wave radar market.

Currently, overseas companies such as Bosch and Continental occupy the main market share of millimeter-wave radar. Domestic companies have accelerated their layout in recent years and have made breakthroughs in 24GHz products. 77GHz products are in the development process.

1. Obvious cost advantage, mainly used in automatic parking systems

Ultrasonic radar is an important auxiliary sensor in the field of autonomous driving. Ultrasonic radars of different frequencies have different sensitivities and detection angles. The current mainstream ultrasonic radars include 40kHz, 48kHz and 58kHz. According to different technical solutions, ultrasonic radars can be divided into analog, four-wire digital, two-wire digital, and three-wire active digital. As the signal anti-interference ability gradually improves, the corresponding technical difficulty and price also gradually increase. According to whether the horizontal and vertical detection angles are the same, ultrasonic radars can also be divided into isotropic and anisotropic.

Mainly used for reversing radar and automatic parking system. There are two types of ultrasonic radars on the market, namely the Ultrasonic Parking Assist (UPA) ultrasonic sensor used to measure obstacles in front and behind the car and the Automatic Parking Assist (APA) ultrasonic sensor used to measure the distance of side obstacles. The reversing radar system is equipped with 4 UPA ultrasonic sensors, and the automatic parking system adds 4 UPA ultrasonic sensors and 4 APA ultrasonic sensors on this basis, forming a 4+4+4 ultrasonic radar layout for the front, side and rear.

The low price is the biggest advantage. Ultrasonic radar uses the time difference between the ultrasonic transmitter sending out the ultrasonic wave and the receiver receiving the ultrasonic wave to measure the distance of the obstacle. The ranging method is relatively simple, and the corresponding ranging range is also relatively limited, and it is impossible to accurately locate the position of the obstacle.

However, compared with millimeter-wave radar and laser radar, ultrasonic radar has a significant price advantage and a lower overall cost, with a single ultrasonic radar selling for about RMB 10-100. Low technical barriers and low costs have created good basic conditions for the promotion and popularization of ultrasonic radar.

The industry scale will maintain steady growth. The popularity of autonomous driving and the increase in the loading rate of ultrasonic radar in reversing radar and automatic parking systems have promoted the development of the automotive ultrasonic radar market. According to P&S Intelligence data, the global automotive ultrasonic radar market size exceeded US$3 billion in 2019, with a steady market growth rate. It is expected that the global automotive ultrasonic radar market size will have a CAGR of 5.1% from 2020 to 2030, and the global market size will reach US$6.1 billion by 2030.

According to TouBao Automotive Research Institute, the scale of China's ultrasonic radar market has continued to expand since 2014, and it is expected that by 2023, the scale of China's ultrasonic radar market will exceed 6 billion yuan. With the increase in market demand for automatic parking and the replacement of car models, the number of cars equipped with ultrasonic radars is increasing. According to AI Garage data, the penetration rate of reversing radar functions in current domestic models on sale is close to 100%, and the penetration rate of automatic parking systems reached 22% in 2019, and is expected to reach 50% in 2025.

(II) Foreign companies occupy the main market, and domestic companies have mature technologies

Overseas companies such as Bosch and Valeo occupy the main market of ultrasonic radar. According to Hengzhou Bozhi data, in 2018, the top two global ultrasonic radar market shares were Valeo and Bosch, with a combined market share of more than 50%.

Valeo has extensive experience in mass production of ultrasonic radars. Its latest generation of automatic parking system, Park4U, has two modes: parallel and corner. It can complete automatic parking in a small space. It currently provides related products and services to many automakers such as Land Rover, Kia, Volkswagen, and Touran.

Bosch has a series of product lines including ultrasonic radar, reversing radar, semi-automatic parking, and fully automatic parking. Its sixth-generation ultrasonic radar combines digital signal processing with signal coding, and has made great progress in recognition range and recognition speed. It can recognize low objects that the fifth-generation product cannot recognize. Through the cooperation of ultrasonic radar with on-board cameras and millimeter-wave radar, it strives to achieve a more complete L2 autonomous driving system.

Domestic enterprises have mature technologies, and the main barrier is the certification of car companies. Compared with millimeter-wave radar and laser radar, the gap between domestic and foreign manufacturers' ultrasonic radar technology is relatively small. The ultrasonic radars independently produced by a series of domestic enterprises are sufficient to meet the functions of automatic parking and reversing.

However, the supply of ADAS parts, including ultrasonic radar, is currently monopolized by foreign giants. To ensure the smooth development of automobiles, mainstream car companies have fixed suppliers. It is difficult for Chinese independent enterprises to obtain car company certification, and the market space they can enter is small. On the other hand, some independent manufacturers are less competitive in core technologies, have low product added value, and find it difficult to provide complete autonomous driving assistance solutions. Therefore, they are often limited to the second-tier supply echelon with limited profit margins.

Audiway is a leading ultrasonic radar manufacturer in China. Its products are widely used in automotive electronics, smart meters and other fields. Among them, the vehicle-mounted UPA ultrasonic sensor is one of its main products. In 2017, Audiway's ultrasonic sensor market share in China was nearly 30%, and the global automotive reversing radar sensor market share was about 9.59%.

We believe that ultrasonic radar is an important auxiliary sensor for autonomous driving. In 2019, the global automotive ultrasonic radar market size exceeded US$3 billion and will grow to US$6.1 billion by 2030.

Sensors are the key to autonomous driving. The current mainstream sensors mainly include millimeter-wave radar, vehicle-mounted cameras, and ultrasonic radar. In 2020, the penetration rate of L2 autonomous driving in China has reached nearly 15%.

Car companies have launched autonomous driving models with L3 functions. As the level of autonomous driving increases, higher requirements are placed on the quantity and quality of sensors. The number of sensors for L2 autonomous driving is about 6, and for L3 it is about 13. In the future, L5 will have to reach more than 30, which will drive the rapid growth of the automotive sensor market. In 2016, the market size of China's automotive sensor industry reached 10 billion yuan, and it is expected that by 2025, the market size will exceed 60 billion yuan.

In-vehicle cameras are relatively mature and are the most widely used "eyes of the car". In-vehicle cameras are subject to more stringent environmental conditions and have correspondingly higher value. According to ICVTank's forecast, by 2025, the global in-vehicle camera industry will reach US$27 billion, and the Chinese in-vehicle camera industry market will reach RMB 23 billion.

In terms of automotive camera lenses, Sunny Optical is the global leader, and Lianchuang Electronics has also begun supplying leading companies; in terms of CMOS sensors, Will Semiconductor ranks second in global market share and is expected to continue to increase its share in the future; in terms of module and system integration, overseas companies dominate, and domestic companies such as OFILM, Q Technology, Desay SV, HUAYU Automotive, and Huayang Group are actively making plans and are expected to occupy a larger share in the future by virtue of their cost advantages.

Millimeter-wave radar has the advantages of small size, high cost performance, and all-weather operation, and is the core sensor for autonomous driving. From the perspective of technological development, as technology matures and costs decrease, 77GHz radar will gradually replace 24GHz radar and become the mainstream of the future millimeter-wave radar market. In terms of market size, it is estimated that by 2022, the global millimeter-wave radar market will reach US$16 billion, of which short- and medium-range millimeter-wave radar will account for approximately US$8.4 billion and long-range millimeter-wave radar will account for US$7.6 billion.

According to OFweek's forecast, China's millimeter-wave radar market will exceed 30 billion yuan by 2025. At present, overseas companies such as Bosch and Continental occupy the main market share of millimeter-wave radar. Domestic companies have accelerated their layout in recent years. Desay SV and Huayu Automotive have made breakthroughs in 24GHz products, and 77GHz products are in the process of research and development. Huayang Group also has integrated millimeter-wave radar products.

Ultrasonic radar is an important auxiliary sensor for autonomous driving. Ultrasonic radar has obvious cost advantages and is mainly used for reversing radar and automatic parking systems. In 2019, the global automotive ultrasonic radar market size exceeded US$3 billion and will grow to US$6.1 billion by 2030. Overseas companies such as Bosch and Valeo occupy the main market of ultrasonic radar.

(Click on the picture to view the conference agenda)