Due to the surge in popularity of autonomous driving, 4D imaging radar is becoming the forefront of innovation.
On one hand, Tesla, which boldly claimed in 2019 to achieve Full Self-Driving (FSD) using only visual technology, announced at the end of last year that it would reintroduce a 4D millimeter-wave radar into its 4th generation self-driving platform. On the other hand, as a potential competitor to LiDAR, the performance of 4D imaging radar is improving, offering a higher cost-performance ratio while further compressing the space for LiDAR.
Despite ongoing debates surrounding this technology, major OEMs, Tier 1 suppliers, and chip companies are unabashedly innovating around 4D imaging radar. According to Yole's statistics, in 2022, the market size for 4D imaging radar was only $200 million. By 2028, this figure is expected to increase to $2.2 billion, with a compound annual growth rate of 49% during this period.
Advertisement
It can be seen that 4D imaging radar has great potential.
4D Imaging Radar, Not Inferior to LiDAR
In fact, millimeter-wave radar is not a new technology; it was developed and applied in the 1940s and 1950s. However, over the past 20 years, the millimeter-wave radar we referred to was the traditional 3D radar capable of measuring distance, speed, and horizontal angle. In general Advanced Driver Assistance Systems (ADAS) applications, such radar applications were sufficient.
But now, with the accelerated popularization of autonomous driving, the inherent shortcomings of traditional radar, such as poor resolution, low pedestrian reflectivity, noise, and the inability to detect stationary objects (such as roads, fences, signs, etc.), have been magnified. However, some inherent characteristics of millimeter-wave radar make them a choice for automotive intelligence solutions.
To this end, the entire industry has put forward new requirements for millimeter-wave radar.
For example, the original radar only needed to detect a distance of about 100 meters, but now many projects require 150 meters, and the front radar even requires more than 300 meters; at the same time, when the radar is required to see far, weak targets at close range also need to be well detected; in terms of point cloud numbers, in the past, the general radar required about 300 point cloud numbers per frame, but now, this data needs to reach 1000 or even 2000.To address the aforementioned issues, 4D imaging radar—a radar that builds upon traditional radar by adding a vertical angle detection, enhancing output point density, and possessing a higher resolution capability—has quickly gained popularity.
Because of significant improvements in resolution, detection range, and point cloud count, 4D imaging radar can detect data such as relative velocity, distance, azimuth, and altitude. At the same time, 4D imaging radar also has a sufficiently high dynamic range to distinguish smaller obstacles at a distance. It can then identify road signs, static objects, and objects at a greater distance, which traditional millimeter-wave radar cannot do.
Dr. Jia Shu, founder and CEO of Galore Microelectronics Technology Co., Ltd. (hereinafter referred to as "Galore"), also told Semiconductor Industry Observation that the difference between 4D imaging radar and traditional ADAS radar is mainly reflected in two aspects: first, the angular resolution capability is more than double that of the latter, and second, the number of point clouds is also an order of magnitude higher than that of the latter. "Therefore, 4D imaging radar requires more MIMO channels, stronger computing power, and more storage resources," Dr. Jia Shu emphasized. He further pointed out that in many scenarios, 4D imaging radar even has several advantages that LIDAR does not have.
"Firstly, as an optical radar, LIDAR optical sensors, like human eyes, have inherent shortcomings, which is that they do not work in rainy, foggy weather, strong light weather, or under interference conditions. However, millimeter-wave radar is the only sensor that is all-weather and unaffected by weather or light; secondly, millimeter-wave radar is more accurate in measuring speed and has an advantage over optical sensors in handling complex scenarios; moreover, the popularization of active safety, including the recent U.S. NHTSA's introduction of the AEB system as a standard for all vehicles by 2029, requires the cooperation of millimeter-wave radar; in addition, in terms of cost, LIDAR is indeed becoming cheaper and cheaper now. But no matter how much the cost of LIDAR is reduced, it will still be several times more expensive than millimeter-wave radar, and there will be a gap of 5 to 10 times between the two," Dr. Jia Shu continued.
It is worth mentioning that in the middle of last year, the world's leading Tier 1 supplier Bosch announced its withdrawal from the development of high-end autonomous driving LIDAR sensors and reallocated resources to millimeter-wave radar and other sensing technologies, which is enough to prove the great potential of millimeter-wave radar technology.
In the era of 4D imaging radar, these advantages are particularly evident. Especially after millimeter-wave radar has entered the CMOS manufacturing era from the past gallium arsenide and silicon germanium manufacturing, suppliers can provide higher integration, better performance, and more cost-effective solutions, further consolidating the position of 4D imaging radar in intelligent driving.
The huge market prospects have attracted both traditional and emerging millimeter-wave radar newcomers to enter this market.Fierce competition, domestic manufacturers break through
We have to admit that, like many other chips, the millimeter wave radar market has always been a market monopolized by overseas giants.
According to Yole statistics, the six major companies of Continental, Bosch, Hella, Aptiv, Denso, and Veoneer control the millimeter wave radar market. In terms of chips, the $1.8 billion market is mainly divided among manufacturers such as NXP and Infineon. After entering the 4D imaging radar market, the overall competitive landscape has not changed much. The leading chip manufacturers in millimeter waves are basically still the leading suppliers of 4D imaging radar chips, but many new competitors have emerged.
For example, Renesas expanded its influence in the radar market by acquiring Steradian; Arbe launched a 4D millimeter wave radar product for vehicles as early as 2019; Mobileye, which is good at making ADAS chips, has also stood out in this field; Altos Radar launched its breakthrough 4D imaging radar at the 2024 CES. Recently, Gaotian also launched a dual-chip level-linked imaging radar solution based on the Andes platform, joining the increasingly hot 4D imaging radar market.
In the current era of the rapid rise of new energy vehicles, the release of Gaotian's solution is particularly important.
"In terms of products, we compete with competitors in some differentiated competition. By differentiating the architecture and innovation to achieve a leap in radar system performance." Gaotian's technical director Liu Hongquan said when talking about the company's layout in the 4D imaging radar market. "Compared with the current market solutions, the Andes solution basically represents the forefront of millimeter wave radar technology, with many powerful functions, which can achieve 4D high-end radar and imaging radar functions." Liu Hongquan added.
From a technical point of view, the innovative solution of 4D imaging radar aims to improve resolution and other performance indicators to meet the requirements of ADAS/AD applications. To achieve higher resolution, a larger aperture is needed, which is related to the number of transmission and reception channels and the design of the antenna array. Therefore, in the RF module, the Andes solution includes a 7-bit phase shifter based on transmission line design to achieve better MIMO performance, optimized layout, which helps to achieve better isolation; a multi-level waveform loading architecture can support flexible generation of various waveforms, precise digital compensation units, and other functions; integrated JTAG, Aurora, various interfaces, multiple data channels, and analysis modules, which can make the solution more flexible during debugging.
For 4D imaging radar, there is another focus, that is, how to integrate more TX/RX channels in a single system to improve its resolution. To address this issue, Gaotian introduced a flexible cascading technology (Flex-Cascading) based on the self-developed C2C (chip to chip) interface in the Andes solution, which can flexibly cascade multiple chips, potentially reducing the complexity of hardware design and optimizing the overall size and power consumption of the radar sensor. In terms of software development, developers can also reuse the same software as when a single chip is used when multiple chips are cascaded, which further reduces the development cost for developers. It is worth mentioning that these cascaded SoCs can not only synchronize the RF analog to form a larger transmission and reception array but also achieve data exchange and synchronization.
To cope with the increased processing requirements brought about by the large amount of data, Gaotian has provided Andes with more powerful performance and computing power, which not only includes the RSP (Radar Signal Processor) that can achieve ultra-fast processing of radar signals, with a computing speed of up to 23G MACs of the DSP core, but also provides a quad-core CPU with more than 2500 DMIPS computing power, making Andes more at ease when processing data.The RSP of Andes can also change the data flow through instructions, modify the accelerator configuration with software code, implement new algorithms, and increase the flexibility of the solution. This is a stark contrast to the fixed data flow design of BBA in the past, which is in line with the current trend of software-defined vehicles, Liu Hongquan emphasized.
At the same time, in terms of network security, which is of great concern to customers and consumers, the Andes chip adopts a "security island" design, supporting hardware accelerators for commonly used encryption and decryption algorithms, providing technical protection for user privacy and network security. In addition, in response to the development complexity, long landing cycle, and high cost of the 4D imaging radar solution, Gattlan also supports a "Ready-to-deploy" turnkey solution, providing convenience for accelerating the widespread application of 4D imaging radar in smart cars.
"The improvement of system integration will bring challenges in terms of reliability, and the design difficulty will also increase. There are not many companies that can achieve this level, which is also one of Gattlan's strengths and our technical advantages," Liu Hongquan summarized. "This also makes the company one of the few suppliers in the world that can provide SoC single-chip solutions." Liu Hongquan continued.
The accumulation of the past ten years has allowed Gattlan to ship a total of 8 million radar chips by Q1 of this year, and the company's products have also entered more than 20 car companies, achieving more than 200 models equipped. Looking forward to the next ten years, Gattlan is determined to become a global leader in millimeter wave radar chips and promote millimeter wave radar to become a truly inclusive technology.
Towards this goal, on the one hand, Gattlan launches products for applications in automotive active safety. At the same time, Gattlan will also make product layouts for applications in smart homes, smart elderly care, industrial automation, and smart cities, aiming to better serve people's daily lives with millimeter wave perception technology.
In conclusion, although it is unanimously optimistic, from the current status of the industry, 4D imaging radar still faces significant challenges.
For example, to improve the resolution of the radar, we can choose to increase its operating frequency. However, due to regulatory restrictions on radar frequency, this is not something that can be easily changed. Of course, we can also choose to increase the aperture size to obtain higher resolution, but such a design is likely to bring challenges such as excessive product size, difficulty in heat dissipation, and difficulty in integration on bumpers. The accompanying software complexity is another problem that 4D imaging radar players have to face.
Superimposed on the high specifications required by the application itself, the popularization of 4D imaging radar in the automotive market will not be a simple matter.In Dr. Chen Jiashu's view, under the demand for L2+ deployment, coupled with the trend in the market where some car manufacturers hope to replace LiDAR with 4D imaging radar, the 4D imaging radar is bound to have a huge growth potential in the future. To this end, suppliers need to further reduce the cost of the radar, while also improving the detection capabilities of the 4D imaging radar.
This is not only the development direction of Gatland but also the common goal of all 4D imaging radars.
Comments