4D Imaging Millimeter-Wave Radar: Industry Pain Points, Technical Breakthroughs and Market Prospects 2026

Last Saturday (March 14, 2026), the second half of the discussion on 4D imaging millimeter-wave radar took place (Background: Hot issues regarding 4D imaging millimeter-wave radar, the answers might all be here).

Several experts in autonomous driving, radar system architecture, RF chips, and radar algorithms were invited to continue discussing hot topics such as the current industry status, pain points in automotive applications, technological development directions, and the industrial chain ecosystem, analyzing the real dilemmas and opportunities in the industry.

Below are some of the main conclusions from the second half of the discussion for your reference. (Note: The content only represents the opinions of the experts in this discussion and may not be entirely accurate.)

01. Industry Development

Question 1: What is the market prospect of automotive millimeter-wave radar?

During the discussion, there were different views on the existential value and market space of millimeter-wave radar.

Optimists:

1. The essential foundation for low-level autonomous driving: For low-level autonomous driving systems of L2 and below, millimeter-wave radar is a crucial core sensor. Basic functions such as blind-spot monitoring, ACC (Adaptive Cruise Control), and AEB (Autonomous Emergency Braking) rely on corner radars and forward radars combined with forward-facing vision, which are already highly mature. These vehicle models account for a high proportion in the market, and the demand for radar will not disappear in the short term.
2. Irreplaceable safety redundancy value for high-level autonomous driving: L3/L4 autonomous driving has stringent requirements for functional safety and SOTIF (Safety of the Intended Functionality). In extreme scenarios where vision and LiDAR are prone to failure, such as dense fog, blinding strong/dim light, and tunnels, millimeter-wave radar is the core guarantee for the vehicle to achieve fallback functions like deceleration and safe pull-over, making it a rigid requirement for high-level autonomous driving access.
3. Promotion by mandatory autonomous driving regulations: Regulations are a significant driver for the current millimeter-wave radar market. Vision-based solutions have a high probability of failure when dealing with small obstacles and tunnel accidents, elevating the importance of 4D imaging millimeter-wave radar.

Pessimists:

1. OEMs reducing configurations and low recognition of radar value: Some OEMs are gradually reducing the number of millimeter-wave radars, from 5 radars to 3, and then to a single front radar. The autonomous driving algorithm teams have a low utilization rate of millimeter-wave radar data, and the weight of radar data in the algorithm pipeline is minimal. Many OEMs' core demand for radar remains solely longitudinal velocity detection, and the system value brought by its performance improvement is not fully recognized.
2. The industry falls into a vicious cycle of cost reduction: Industry price wars have led to meager corporate profits and a severe lack of R&D investment. Many companies have shifted their R&D focus to cost-reduction solutions rather than cutting-edge technologies and performance improvements. With insufficient motivation for technological innovation, the industry is trapped in a vicious cycle of "the more prices drop, the less money for R&D; the less R&D, the more they can only compete by lowering prices." While radar sales grow, the industry's overall revenue and profits continue to decline, making it difficult for companies to invest in R&D, which in turn hinders technological iteration. Leading suppliers can barely maintain themselves through scale, while small and medium-sized manufacturers face survival crises.
3. Severe brain drain and technological generational gap: Multiple experts mentioned that the brain drain in the millimeter-wave radar industry is very severe. On the one hand, salaries and benefits are not as good as those in the Internet or AI fields; on the other hand, despite high technical difficulty and a long industrial chain, the industry fails to gain sufficient attention from OEMs, leading to a lack of sense of achievement among practitioners. The loss of the middle tier (young and middle-aged professionals) is particularly severe. Many have switched to LiDAR or sensor fusion perception, leaving only highly senior veteran experts and newly graduated entrants, forming a talent gap.

Question 2: What are the core pain points faced by automotive 4D imaging millimeter-wave radar?

1. Unstable targets and high false alarm rate: Traditional radar signal processing is based on ideal assumptions such as sparse point targets, uniform motion, and far-field conditions. However, in automotive scenarios, vehicles and obstacles are surface targets with complex motion states, which easily leads to high false alarm rates, numerous ghost targets, and unstable target tracking.
2. Radar aperture bottleneck caused by spatial constraints: The installation space in vehicles limits the radar array aperture. Simply increasing the number of transceiver channels has limited improvement on physical angular resolution. Moreover, the angular resolution directly ahead claimed by the industry differs greatly from the actual performance in wide-angle scenarios such as 45°. Wide-angle scenarios suffer from high side lobes and prominent false alarm issues, causing a disconnect between paper parameters and actual vehicle performance.
3. High difficulty in algorithm pipeline adaptation: The signal characteristics and data formats of millimeter-wave radar differ greatly from those of vision and LiDAR, making it difficult to integrate into the current mainstream end-to-end autonomous driving algorithm pipelines. OEM perception algorithm teams have insufficient understanding of radar data, leading to extremely low utilization rates. In most cases, it is only used as a fallback backup, or even completely unused.
4. Poor robustness in target tracking: Traditional radar target tracking algorithms heavily rely on a massive amount of if-else patches to adapt to automotive scenarios. In complex scenarios such as curves, tunnels, and multi-vehicle intersections, tracking crashes and false braking caused by false targets are prone to occur. In some cases, radar data completely contradicts vision and LiDAR data, leading to system decision failures.

02. Technical Research

Question 3: What are the advantages and landing bottlenecks of the satellite radar solution?

The satellite radar is essentially an architectural reconstruction of millimeter-wave radar. It migrates all signal processing and algorithm computations, originally executed by the local MCU (Microcontroller Unit) at the radar front end, to the vehicle domain controller. The radar body only retains the antenna, RF, ADC acquisition, and SerDes serial transmission modules. By removing the local processor, it transmits FFT data or even raw ADC data to the domain controller via in-vehicle Ethernet/SerDes, where the domain controller completes the subsequent full-link signal processing, target detection, and tracking.

Core Advantages:

1. Significant cost reduction: Removing the local radar processor substantially reduces the hardware cost of the radar body. It is the core path for the industry to achieve a price level of around 100 RMB for 4D radars, and also the value most valued by OEMs.
2. More flexible algorithm upgrades and fusion: Leveraging the computing power of the domain controller, it can run complex algorithms such as super-resolution algorithms and deep learning filtering that the local MCU cannot implement. Meanwhile, algorithm iterations do not require radar hardware upgrades, making algorithm debugging and OTA easier.
3. Increased hardware standardization: The radar body is simplified to a pure hardware front end, lowering the R&D threshold for radar manufacturers. It also facilitates OEMs in achieving multi-supplier substitution and further controlling the supply chain.

Landing Bottlenecks:

1. High computing power and memory consumption: Processing raw ADC data from a single radar may require 10 TOPS of general computing power and simultaneously occupy 7-10GB of DDR memory (information provided by participating experts, for reference only). However, domain controller resources of OEMs are prioritized for vision modules, making it difficult to allocate sufficient resources for radar processing. This is the biggest obstacle to its mass production and landing.
2. Decreased system real-time performance: Traditional radar local processing has extremely low latency, whereas the satellite solution needs to wait for the entire frame of data transmission to be completed before processing, which brings additional system latency.
3. Limited control capability of the radar front end: Functions such as anti-jamming and adaptive waveform adjustment require fast closed-loop feedback from the radar front end. The transmission delay of the satellite solution makes such real-time control difficult to achieve, affecting the radar's performance in complex scenarios.

Question 4: Is distributed coherent synthetic radar a technological breakthrough direction for the industry? What are the core challenges?

Distributed coherent synthetic radar is an important research direction for millimeter-wave radar to break through the upper limit of physical performance. Theoretically, it has great potential for performance improvement, but the difficulty of engineering implementation is extremely high, making it difficult to achieve automotive mass production in the short term.

Theoretical and Technical Value:

1. Breaking the bottleneck of physical angular resolution: Through the distributed array layout of multiple corner radars on the vehicle body, a super-large array aperture of the meter level can be synthesized, achieving an order-of-magnitude improvement in angular resolution at the physical level and realizing ultra-high angular resolution.
2. Significantly improving target detection stability: Multi-node and multi-angle observation perspectives can greatly suppress the RCS scintillation problem of millimeter-wave radar, making target tracking more robust and solving false alarms and ghost targets from the underlying level.
3. Qualitative leap in anti-jamming capability: It can achieve focused suppression of interference sources in three-dimensional space, fundamentally solving the industry pain point of mutual interference among automotive radars, which is a capability unachievable by single-radar solutions.

Core Landing Challenges:

1. Extreme difficulty in phase synchronization: Coherent synthesis of distributed radars requires carrier-wave-level high-precision phase synchronization. Military solutions can achieve synchronization via wireless signals, whereas automotive scenarios can only rely on hardwire connections. Currently, only the feasibility of low-frequency PLL synchronization has been verified in laboratories, and low-cost, automotive-grade synchronization solutions have not yet been achieved.
2. Complete reconstruction of the signal model required: Under ultra-large apertures, the far-field angle measurement model of traditional radars completely fails. It is necessary to reconstruct the entire signal processing framework for near-field imaging and wavefront focusing. The industry lacks sufficient relevant technical accumulation.
3. Steep increase in mass production calibration difficulty: Process deviations in automotive installation, vehicle vibration, and temperature changes will all lead to array phase deviations, requiring real-time, high-precision array calibration, which is almost impossible to achieve in large-scale mass production scenarios.
4. High difficulty in vehicle architecture transformation: This solution requires reconstructing the hardware architecture, transmission links, and computing power allocation of the entire vehicle's radars. It involves massive transformation of the vehicle's E/E (Electrical/Electronic) architecture, making it highly difficult to promote among OEMs.

Question 5: What other frontier technology exploration directions are there for 4D imaging millimeter-wave radar?

During the meeting, some experts proposed several possible frontier exploration directions in the industry, such as:

1. Metasurface radar technology: Through reconfigurable metasurface antennas, waveform modulation is directly completed at the hardware level, replacing the traditional dozens of transceiver channels. This can greatly simplify the system architecture and reduce performance instability and calibration difficulties caused by multi-channel heating. However, this technology has a narrow operating bandwidth and high engineering difficulty. Previously, the Silicon Valley startup Metawave tried it and went bankrupt. Currently, it is mainly applied in the 6G communication field, and automotive landing still requires long-term technical accumulation.
2. Algorithm innovation: Traditional radar detection relies on CFAR (Constant False Alarm Rate) detection. Solutions based on energy thresholds have poor adaptability to complex scenarios. The industry has begun to explore innovation at the signal representation level, drawing on the signal processing methods of gravitational wave detection to optimize weak signal detection capabilities, solving the mismatch between traditional models and automotive scenarios, and reducing the false alarm rate from the underlying level. Additionally, the industry is applying deep learning and neural networks to radar CFAR detection, point cloud denoising, target tracking, and other links to replace traditional rule-based algorithms, in order to improve detection robustness in complex scenarios.
3. Technical exploration of higher frequency bands: The industry has begun to explore millimeter-wave radars in higher frequency bands such as 120GHz and 140GHz. Higher frequency bands can achieve larger array apertures in smaller sizes, improving angular resolution while significantly reducing antenna volume. However, it faces issues such as frequency band compliance, process difficulty, and limited short-distance measurement due to high atmospheric loss. Currently, only startups have demo displays (e.g., TERADAR), with no mature automotive products.
4. Integrated sensing and communication (ISAC): ISAC refers to using the same set of hardware to simultaneously achieve radar detection and communication, such as vehicles directly communicating and perceiving the surrounding environment through millimeter-wave frequency bands. This idea is theoretically beautiful but faces challenges such as spectrum resources, protocol standardization, and cost. Currently, UWB (Ultra-Wideband) technology has been attempted to replace ultrasonic waves in areas like car keys and parking assistance, but its use for vehicle-to-vehicle communication remains distant. One expert believes that ISAC may be more promising in drone swarm scenarios because protocols can be unified and it is less sensitive to cost.

03. Non-Automotive Landing Scenarios

Question 6: Does 4D imaging millimeter-wave radar have landing potential in drone scenarios?

During the meeting, an expert proposed the idea of applying millimeter-wave radar for obstacle avoidance in drones. Current visual obstacle avoidance solutions for consumer drones have obvious limitations in high-speed flight, low-light environments, and complex scenarios. Millimeter-wave radar can achieve longer-distance detection, has stronger recognition capabilities for small targets like wires, and is unaffected by light and weather, making it an ideal supplementary solution for high-speed obstacle avoidance in drones.

Industrial drones for agricultural plant protection and power line inspection have begun to apply millimeter-wave radar for wire detection, terrain mapping, obstacle avoidance, and other scenarios, verifying technical feasibility. There is huge potential demand in the consumer drone market.

Core Landing Challenges:

1. Significant volume and installation constraints: Consumer drones have small fuselage space, making it impossible to install traditional planar array radars. Conformal array antennas need to be designed, but issues such as array calibration and performance degradation of conformal arrays have not yet formed mature mass production solutions.
2. Strict power consumption constraints: The endurance of consumer drones is only 30-40 minutes. The power consumption of traditional automotive millimeter-wave radars would significantly compress their endurance time. Low-power design is the core threshold for its large-scale landing.
3. Higher cost requirements: Consumer drones are more sensitive to hardware costs than automotive scenarios. Automotive radar solutions at the 100 RMB level still cannot meet their needs. It needs to be reduced to the around-10-RMB level to have the possibility of large-scale landing.
4. Difficult to break through the industry ecosystem closed loop: Leading manufacturers like DJI have formed their own self-developed sensor solution systems. It is difficult for external radar manufacturers to enter, and the industry lacks large-scale landing scenarios.

Question 7: What other non-automotive landing scenarios are there for millimeter-wave radar?

1. Smart home and human perception: Leveraging the privacy protection advantages of millimeter-wave radar, it achieves human presence detection, fall detection, and vital sign monitoring, replacing cameras and infrared sensors. It has achieved small-scale landing in smart switches and smart lights. Meanwhile, driven by mandatory overseas regulations, the in-cabin occupant monitoring scenario has achieved large-scale installation, becoming the most stable incremental market for non-forward radars currently.
2. Overseas market for two-wheelers and low-speed vehicles: In the electric motorcycle and low-speed mobility vehicle market, there is a clear demand for AEB and blind-spot monitoring functions of millimeter-wave radar. The annual shipment volume in related scenarios can reach hundreds of thousands of sets, which is the core incremental market for domestic millimeter-wave radar manufacturers going overseas currently.
3. Industrial and security scenarios: For obstacle avoidance in industrial AGV/AMR and perimeter intrusion detection in the security field, millimeter-wave radar can achieve 24-hour uninterrupted detection under strong light, low light, and harsh weather, making up for the shortcomings of cameras. It has been implemented in some special scenarios.
4. Medical health and elderly care: Leveraging the phase detection capability of millimeter-wave radar, it achieves non-contact heart rate, respiration, and heartbeat monitoring, and even cough behavior detection. There is clear demand in elderly care scenarios, with abundant related academic research achievements, but very little engineering implementation.

04. Conclusion

This 4D imaging millimeter-wave radar discussion invited a total of about 20 frontline industry experts, with a total discussion time of about 8 hours. The main conclusions are summarized as follows:

1. The core value of millimeter-wave radar in autonomous driving sensor suites lies in its all-weather adaptability and direct velocity measurement capability. 4D imaging millimeter-wave radar has higher resolution and longer detection range, making it indispensable in meeting mandatory regulations and providing safety redundancy for high-level autonomous driving of L3 and above.
2. False alarms and insufficient angular resolution of radar remain the main technical bottlenecks. Generally, resolution is improved by increasing the number of virtual channels and physical apertures. However, automotive installation space is limited. Under the same physical installation aperture, increasing the number of transceiver channels mainly optimizes side lobe suppression and anti-jamming capabilities, with limited marginal benefits for angular resolution improvement. More technological innovation is needed to fundamentally solve core industry pain points such as multipath ghosts, target scintillation, false alarms, and weak target detection failure.
3. Satellite radar is an important evolutionary direction and the main technical path for current cost reduction and performance improvement. However, this solution faces challenges such as computing resource competition, data communication bandwidth, increased system latency, and limited real-time processing capabilities such as waveform control and anti-jamming at the radar front end. The technical solution is still being continuously optimized.
4. Due to the high difficulty in annotating radar point clouds and constructing high-quality training datasets, deep learning technology has hardly been applied at scale. Only simple machine learning methods such as random forests and multi-layer perceptrons are used in links like target classification and target size regression. End-to-end deep learning radar algorithms are still in the R&D stage.
5. Frontier technologies such as distributed coherent synthetic radar, metasurface antennas, and PMCW (Phase-Modulated Continuous Wave) waveforms can theoretically achieve radar aperture expansion, angular resolution leap, and significant improvement in anti-jamming capabilities, representing potential directions for long-term technological breakthroughs in the industry. However, all these technologies face immense engineering implementation difficulties, remain in the academic research stage, and cannot achieve automotive large-scale mass production in the short term.
6. Domestic millimeter-wave radar chips have made considerable progress, but there is still a certain gap compared to foreign leading manufacturers in terms of the consistency, stability, and reliability of RF performance. This affects the stability and anti-jamming capability of the radar's backend range, velocity, and angle measurement. To maintain market share, foreign giants have spared no effort in slashing prices, forming huge price suppression on domestic chips. It remains highly difficult for domestic MMIC (Monolithic Microwave Integrated Circuit) manufacturers to achieve further breakthroughs in the automotive OEM market.
7. The price war in the automotive millimeter-wave radar industry has entered a white-hot stage. The industry as a whole faces the challenge of "increasing sales but declining revenue." Radar manufacturers can hardly achieve profitability in the core automotive segment and can only exchange low prices for market share. Since cutthroat price competition compresses manufacturers' R&D investment space, the millimeter-wave radar industry is facing severe brain drain and technological generational gaps. Core technical talents are massively transferring to sectors like LiDAR and autonomous driving algorithms. Newly entered talents lack confidence in the long-term development of the industry. The industry lacks technological accumulation and innovation motivation, severely restricting the healthy development of the industry.
8. Under the high-intensity cutthroat price competition in the core automotive segment, segmented markets such as two-wheelers (electric motorcycles, motorcycles), industrial inspection, drones, and smart home/elderly care have become important breakthrough directions for domestic radar manufacturers. However, many of these scenarios face strong competition from low-cost solutions like cameras and UWB, and have high requirements for radar low power consumption and miniaturization. There are still many challenges for large-scale landing.

Xueling believes that,

4D imaging millimeter-wave radar is standing at a critical crossroads where opportunities and challenges intertwine.

As an irreplaceable core perception unit in autonomous driving systems, with the strongest adaptability to extreme environments, it firmly holds the bottom line of safety redundancy for autonomous driving, making it an indispensable core link in the landing process of high-level intelligent driving.

Along with the large-scale landing of satellite radar technology, the performance boundaries of 4D imaging millimeter-wave radar will continue to break through. Relying on the deep excavation of technical value and the continuous deepening of scenario-based landing, 4D imaging millimeter-wave radar will surely occupy a more important position in the evolution of autonomous driving.

We sincerely hope that the industry can break out of the vicious cycle of cutthroat price competition as soon as possible, truly complete the underlying transformation from "cost slaughter" to "value deep cultivation," let technology return to value itself, and jointly achieve a safer and more intelligent new future for autonomous driving.

Once again, thank all the experts and teachers who participated in the discussion of this closed-door meeting on 4D imaging millimeter-wave radar.

We will continue to organize special exchange meetings on satellite radar later. If you are interested, please feel free to contact Xueling. Registration page: https://dcn7get8fskg.feishu.cn/wiki/XfsEwavjMieBdIkJSyBcxqbOnWr

Extended reading: https://www.eefocus.com/article/1967581.html

Personal views, not necessarily accurate, discussions are welcome. I am Xueling, researching AI (Artificial Intelligence) technologies, products, and applications. Welcome to communicate.