시장보고서 | 연간정보서비스 | 맞춤형시장조사 | 메일링 | 샘플 요청 목록
자동차용 4D 레이더 산업(2025년)
Automotive 4D Radar Industry Research Report, 2025
상품코드 : 1892142
리서치사 : ResearchInChina
발행일 : 2025년 12월
페이지 정보 : 영문 260 Pages
 라이선스 & 가격 (부가세 별도)
US $ 3,800 ￦ 5,574,000
Unprintable PDF (Single User License) help
PDF 보고서를 1명만 이용할 수 있는 라이선스입니다. 인쇄 불가능하며, 텍스트의 Copy&Paste도 불가능합니다.
US $ 5,700 ￦ 8,361,000
Printable & Editable PDF (Enterprise-wide License) help
PDF 보고서를 동일 기업의 모든 분이 이용할 수 있는 라이선스입니다. 인쇄 가능하며 인쇄물의 이용 범위는 PDF 이용 범위와 동일합니다.


한글목차

1. 4D 이미징 그레이더는 "옵션"에서 "필수"의 센서로 변모했습니다.

4D 레이더는 거리, 속도, 방위, 고도 외에도 물체 높이 데이터의 감지 및 분석 기능을 추가합니다. 날씨와 조명 조건의 영향을 받지 않기 때문에 자동 운전 시스템에 필수적인 센서가 되고 있습니다. 그 개발은 주로 다음 요인에 의해 추진되고 있습니다.

1. 정책. 2025년 4월, National Technical Committee of Auto Standardization(NTCAS)은 "경차용 자동 긴급 브레이크 시스템의 기술 요건과 시험 방법(초안)"을 발표하여 기존 AEB 시스템에 권장된 국가 표준, GB/T 39901-2021을 대체했습니다. 이 규정에서는 AEB 시스템을 「임의 장비」에서 「표준 장비」로 단계적으로 이행시키는 것을 제안하고, 2028년 1월 1일 이후, M1/N1 차량에는 AEB 시스템의 표준 장비를 의무화할 것을 요구하고 있습니다. 관련 AEB 규제가 제동 전 차량 최고 속도에 대해 더 엄격한 요구를 부과하는 동안 레이더는 더욱 엄격한 성능 기준을 요구하고 있습니다. 구체적으로, 전방 감시 시스템은 보다 긴 감지 거리, 보다 강력한 미약한 물체 인식 능력과 다중 물체 분해능, 그리고 보다 정확한 장애물의 높이 측정이 요구됩니다.

2. 성과의 비약적인 향상은 지각의 약점을 보충합니다. 4D 이미징 그레이더는 기존 레이더에서 인식할 수 없었던 고도의 장애물(높이 제한 극이나 도로 표지 등)과 정지 물체(경사로 위의 불법 주차 차량 등)의 감지 문제를 해결합니다. 각도 분해능은 1-2°(8-32채널 LiDAR 상당)로 향상되고, 점군 밀도는 종래식 레이더(예 : SINPRO SFR-2K의 1프레임당 2048점)의 8배 초과를 실현합니다. 전방 차량에 차단된 물체(예를 들어, 전주차의 브레이크 부분 등)의 윤곽을 선명하게 복원하여 정확한 모니터링을 실현합니다. 동시에 모든 날씨에 대응한 방해파 대책 기능을 갖추고 비, 눈, 안개, 모야 등의 가혹한 환경 하에서도 검지 거리는 300미터에 달하고, 카메라와 LiDAR을 대폭 웃도는 성능을 발휘합니다.

3. 비용 우위성이 대규모 채용의 핵심적인 촉진요인이 되었습니다. 듀얼 캐스케이드 시스템이 쿼드 캐스케이드 시스템이나 단일 칩 집적화를 대체하는 경향이 있기 때문에 4D 기본 이미징 그레이더의 단가는 2024년 초 500-1,000위안에서 2025년에 200-400위안으로 저하해, 종래식 레이더의 가격대에 접근함과 동시에, LiDAR의 1/5-1/10의 수준이 되었습니다(LiDAR의 청소 코스트는 레이더 본체와 동등합니다).

2. 2030년까지 4D 레이더 점유율은 50%를 초과할 것으로 예측됩니다.

ResearchInChina의 예측에 의하면, 2024년·2025년의 4D 레이더 센서 탑재수는 각각 273만 7,000대, 1,106만대가 되었습니다. 2030년까지 5,000만대를 돌파해 보급률은 2025년의 26.0%에서 54.5%로 상승할 것으로 예측되고 있습니다. 이에 따라 전방 4D 레이더와 4D 코너 레이더의 보급률도 급상승하고 특히 4D 코너 레이더가 가장 급속한 성장을 보여줄 전망입니다.

제품 선정에서 OEM은 4D 레이더를 카메라와 LiDAR을 보완하는 중요한 기술로 자리 매김하고 있습니다. 카메라가 고해상도의 의미론적 이해와 색 정보를 담당하고, LiDAR가 치밀한 3D 형상을 제공하는 한편, 4D 레이더는 시야 불량시나 복잡한 전자기 환경하에서도 안정된 거리·속도·고도 정보를 제공합니다. OEM은 성능과 비용의 균형과 통합 지각과 같은 요소를 고려합니다.

통합 지각이 주류가 되고 OEM은 지각 하드웨어의 확충·고도화를 적극적으로 진행하고 있습니다. OEM(예 : ONVO L60)은 기본적으로 "4D 레이더 비전"솔루션을 채택하고 하이 엔드 모델(예 : Maextro S800)은 LiDAR을 추가하여 중복성을 보장합니다. 알고리즘 수준에서 BEV Transformer 아키텍처는 다중 센서 퓨전의 표준 솔루션으로 시간 모델링을 통해 물체 추적의 안정성을 향상시킵니다.

본 보고서에서는 중국의 자동차용 4D 레이더 시장에 대해 조사했으며, 시장 규모와 탑재수 예측, 가격대 분석, 국내외 기업 정보 등을 제공합니다.

목차

용어

제1장 자동차용 4D 레이더의 개요

제2장 자동차용 4D 레이더 시장

제3장 중국의 자동차용 4D 레이더 기업

제4장 해외의 자동차용 4D 레이더 기업

제5장 자동차용 4D 레이더 칩/안테나 기업

제6장 자동차용 4D 레이더의 요약과 동향

KTH
영문 목차

영문목차

4D radar research: From "optional" to "essential," 4D radar's share will exceed 50% by 2030.

1. 4D imaging radar has transformed from an "optional" to a "must-have" sensor.

4D radar adds the detection and analysis of object height data, perceiving distance, speed, azimuth, and altitude. It is immune to weather and lighting conditions as an indispensable sensor for autonomous driving systems. Its development is mainly driven by the following factors:

1. Policies. In April 2025, the National Technical Committee of Auto Standardization (NTCAS) of China released the "Technical Requirements and Test Methods for Automatic Emergency Braking Systems of Light-Duty Vehicles (Draft)" to replace the original recommended national standard GB/T39901-2021 for AEB systems. It suggested that AEB systems should gradually move from "optional installation" to "mandatory standard configuration" and required that from January 1, 2028, M1 and N1 vehicles must be equipped with AEB systems as standard. As relevant AEB regulations impose increasingly stringent requirements on the maximum speed of vehicles before braking, radar faces more stringent performance standards: forward-looking perception systems must have longer detection ranges, stronger weak object recognition and multi-object resolution, and more accurate obstacle height measurement.

On September 17, 2025, the Ministry of Industry and Information Technology of China publicly solicited opinions on the mandatory national standard "Safety Requirements for Combined Autonomous Driving Systems of Intelligent Connected Vehicles". This standard strengthens technical supervision in the field of autonomous driving where accidents occur frequently. In its technical draft, it puts forward higher requirements for the perception capabilities of autonomous driving systems in all weather conditions.

With the upgrading of global safety regulations and the increasing penetration rate of L2+/L3 autonomous driving, highway NOA and urban NOA rely on radar, especially 4D imaging radar, to make up for the defects in visual perception and the decline of LiDAR functions (such as in rain, snow, fog, low light, nighttime, severe weather, and obstructions, etc.). For example, when a vehicle is traveling at high speed, the AEB system needs to reliably complete its task. It not only needs to detect large vehicles, but also to recognize smaller, less reflective, or fast-moving objects, such as children crossing the road or motorcycles that have fallen over. Moreover, such detection often occurs in environments with insufficient light or in rain, snow, or fog. There is also the challenge of detecting stationary objects at a distance, such as cardboard boxes, people next to highway guardrails, and construction equipment. Currently, there are solutions in the industry that enable AEB with a single 4D imaging radar sensor, such as the Aumovio ARS620, which can meet the national AEB standard with a single radar sensor and a detection range of 280 meters (cars and motorcycles) and 174 meters (pedestrians).

2. Performance leap makes up for the shortcomings in perception. 4D imaging radar solves the problem that traditional radar cannot recognize high-altitude obstacles (such as height restriction poles and road signs) and static objects (such as illegally parked vehicles on ramps). Its angular resolution is improved to 1-2° (equivalent to the level of 8-32 channel LiDAR), and its point cloud density is more than 8 times that of traditional radar (such as 2048 points per frame of the SINPRO SFR-2K). It can clearly restore the object outline and achieve accurate monitoring of objects obscured by vehicles in front (such as the brakes of vehicles in front). Meanwhile, it has all-weather anti-interference capabilities, and its detection range can still reach 300 meters even in harsh environments such as rain, snow, fog and haze, which is significantly better than cameras and LiDAR.

In reality, 4D radar can be divided into three types. The first type addresses basic altitude perception, with a point cloud density of less than 4,000 points/second and a range within 300 meters. The second type is 4D imaging radar, which provides high-resolution imaging for positioning, with a point cloud density typically between 30,000 and 100,000 points/second, a range within 350 meters, and an elevation angle of 0.8-1°. Examples include Huawei's 4D imaging radar, SINPRO's 4D imaging radar based on satellite architectures, and Arbe Phoenix. The third type is 4D digital imaging radar, which provides intelligent real-time perception for positioning, with a point cloud density typically higher than 100,000 points/second, a range within 400 meters, and an elevation angle improved to 0.5°-0.8°, enabling the detection of small objects and lane lines. The three are in a progressive relationship, jointly promoting the upgrade of perception redundancy in autonomous driving.

3. Cost advantage has become the core driving force for large-scale application. As dual-cascaded systems tend to replace quad-cascaded systems and single-chip integration, the unit price of 4D basic imaging radar has dropped from RMB500-1000 in early 2024 to RMB200-400 in 2025, approaching the price range of traditional radar and only 1/5 to 1/10 of that of LiDAR without requiring an additional cleaning system (the cleaning cost of LiDAR is almost the same as that of the radar itself).

With the optimization of chip processes (such as NXP's S32R47 processor and Milliverse's MVRA188 8-transmitter/8-receiver chip) and economies of scale, the cost is expected to drop below $100, which will accelerate application and popularization. 4D imaging radar has become a must-have option in the era of equal access to autonomous driving safety. 4D imaging radar has been added, or 4D radar has replaced the original traditional radar.

2. By 2030, 4D radar will account for over 50%.

According to ResearchInChina, 2.737 million and 11.06 million 4D radar sensors were installed in 2024 and 2025 respectively. The figure is projected to exceed 50 million by 2030, with the penetration rate rising from 26.0% in 2025 to 54.5%. Correspondingly, the penetration rates of forward-facing 4D radar and 4D corner radar will also jump, with 4D corner radar showing the fastest growth.

In terms of product selection, OEMs regard 4D radar as an important technological supplement to cameras and LiDAR. Cameras are responsible for high-resolution semantic understanding and color information, LiDAR provides dense 3D shape, and 4D radar offers stable distance, speed, and altitude information in low visibility or complex electromagnetic environments. OEMs consider factors such as performance-cost balance and integrated perception.

Integrated perception has become the mainstream, and OEMs are actively increasing and upgrading their perception hardware. OEMs (such as the ONVO L60) generally adopt the basic "4D radar + vision" solution, and high-end models (such as the Maextro S800) add LiDAR to create redundancy. At the algorithm level, the BEV+Transformer architecture has become the standard solution for multi-sensor fusion, improving object tracking stability through temporal modeling.

3. 4D radar develop toward three directions

1. Chip processes continue to evolve towards more advanced levels, with continuous improvement in integration and performance.

As the "heart" of radar, the radio frequency MMICs is the most critical in the industry chain. MMICs have undergone iterative upgrades from GaAs to SiGe and then to CMOS. Because CMOS wafers are inexpensive and highly integrated, a radar only requires one RF front-end MMIC and one BBIC, further reducing the system cost by 40%. For example, NXP's 28 nm RFCMOS radar chip - SAF85xx has significantly improved performance compared to the previous 45 nm product, while its cost has been greatly reduced. Calterah's Andes premium 8T8R imaging radar solution connects two 4T4R Andes SoCs (22nm CMOS radar SoC - Andes RoP chip) via C2C, simplifying the hardware design architecture and making it more competitive in terms of system cost. It can achieve a maximum detection range of 350 meters.

As the core components of 4D radar, RF MMICs and processors account for more than 50% of the cost. Currently, there are different solutions for efficiency improvement and cost reduction in the industry, and players choose different routes.

Chip cascading: Combining multiple MMICs (such as two 3T4R chips forming a 6T8R) increases the number of channels to enlarge the aperture. Its advantages lie in a short development cycle and a mature industrial chain, while its disadvantages include high power consumption, large size and low signal-to-noise ratio. For example, WHST's STA77-6 4D radar uses a dual-chip cascade with 6 transmitters and 8 receivers, achieving a detection range of 300 meters. Its 4D ST77-10 has a dual-chip cascade with 16 transmitters and 16 receivers, a field of view of 120° x 30°, a resolution of 1° (horizontal) x 1.5° (elevation), and a detection range of 350 meters. This 24T24R imaging radar solution is built on NXP's next-generation high-performance MPU (S32R47) and cascaded 8T8R chips. Paired with NXP's 24T24R array waveguide antenna reference design, the solution can achieve imaging-level accuracy with 576 virtual channels, meaning it can accurately recognize scattered small objects 160 meters away.

Chip integration: Typical single-chip high integration solutions (such as 8T8R) come from NXP and ANDAR. For example, ANDAR's ADT7880 single-chip solution integrates 8 transmitters and 8 receivers, supports digital beamforming (DBF) architectures and flexible cascading, significantly reducing system complexity and cost.

2. Packaging technology is developing towards higher integration, driving the miniaturization and integration of radar modules.

Currently, radar packaging technologies include AiP, RoP, LoP/LiP, and RoC. Among them, AiP, such as Calterah's Alps AiP and TI's AWR2944, sacrifices some detection range in exchange for extreme miniaturization, making it suitable for in-cockpit applications. LoP such as TI's AWR2544 and NXP's SAF85xx improves the signal-to-noise ratio by optimizing the signal path. Its principle is to transmit the radio frequency signal directly from the bottom of the package to the external 3D waveguide antenna, requiring only 2 signal conversions (bare die -> package substrate -> waveguide), reducing 4 conversions required by traditional packaging. It is suitable for satellite-based radar (corner radar, door handle radar) and L3+ autonomous driving (high angular resolution required). The innovative RoP, by replacing traditional feeders with radiators, handles the insufficient channel isolation in AiP, while avoiding the mechanical stability risks incurred by LoP, representing a new direction for 4D imaging radar.

Microstrip antennas are gradually being upgraded to 3D waveguide antennas. For example, the 8T8R 4D imaging satellite radar single-chip solution from Guibu Microelectronics uses a 3D waveguide antenna to improve the signal-to-noise ratio and transmit/receive isolation, reduce BOM cost by about 30%, and cut down power consumption by about 30% compared to competitors under the same operating conditions. Baolong Technology's high-performance waveguide 4D radar adopts an air waveguide antenna solution, featuring high radiation efficiency, good anti-interference, and active frequency conversion to avoid interference from other automotive radars.

3. Satellite-based 4D radar achieves 'distributed sensing + centralized computing', helping to reduce cost and improve efficiency.

The core of "satellite-based 4D radar" lies in software-hardware decoupling and centralized computing architectures. It separates computing from the sensor and concentrates it in a powerful central domain controller. The radar only retains necessary radio frequency components (such as MMICs and antennas) for data collection, while processing and decision-making are carried out in the domain controller.

On October 30, 2025, SINPRO officially released its next-generation satellite-based 4D imaging radar - 5R system. The system includes a single front-to-center satellite 4D imaging radar sensor (SFR2-4D-S) and four corner satellite 4D radar sensors (SCR2-4D-S).

On October 22, 2025, Chuhang Tech, in collaboration with Guibu Microelectronics, officially released its first 4D satellite-based radar product - forward-facing satellite radar with a single-chip 8T8R integrated waveguide antenna. Its performance is comparable to that of a dual-cascaded 4D radar, while reducing cost by 60%. The integrated waveguide antenna design reduces the size by 30% and improves anti-interference capability by 50%. With "full localization + extreme cost performance", it fills the gap in domestic high-end radar and helps China's automotive industry chain achieve autonomy.

Table of Contents

Terminology

1 Overview of Automotive 4D Radar

2 Automotive 4D Radar Market

3 Chinese Passenger Car 4D Radar Enterprises

4 Foreign Passenger Car 4D Radar Enterprises

5 Automotive 4D Radar Chip/Antenna Enterprises

6 Summary and Trends of Automotive 4D Radar

(주)글로벌인포메이션 02-2025-2992 kr-info@giikorea.co.kr
ⓒ Copyright Global Information, Inc. All rights reserved.
PC버전 보기