L3 autonomous driving is near, and 4D millimeter wave radar is on fire

A few days ago, Beijing issued a notice of unmanned manned demonstration application, allowing the public to provide the public with automatic driving services of "no safety officer in the main driver's seat and safety officer in the co-driver's seat" within 60 square kilometers of the Beijing High-level Autonomous Driving Demonstration Zone. This policy not only makes Beijing the first city in the country to open a pilot project for unmanned operation of passenger cars, but also further promotes the application and landing of autonomous driving technology. However, in addition to the support of the policy level, the realization of automatic driving is also inseparable from the support of the hardware system. As the "unmanned behind the wheel" travel service is gradually and conditionally released, more new technologies will be applied to autonomous vehicles, of which 4D millimeter wave radar is optimistic about more and more manufacturers, and its commercialization process may open the "acceleration" mode.

In 2023, the number of vehicles will exceed one million

4D millimeter-wave radar is not a new technology that is new. In March 2020, Waymo, a subsidiary of Google, released the fifth-generation autonomous driving system perception scheme, upgrading millimeter-wave radar to 4D imaging radar, making 4D millimeter-wave radar technology applied to the vehicle end for the first time. In 2021, Huawei entered the game, launched a high-resolution 4D millimeter wave radar, and plans to achieve mass production in the second half of 2022, pushing the 4D millimeter wave radar to a hot peak.

Recently, as attempts at L3 autonomous passenger vehicles on the road have begun, the attention of 4D millimeter-wave radar has heated up again. Li Tingwei, senior vice president of NXP Global and chairman of Greater China, said that the current penetration rate and commercialization of autonomous driving in China are accelerating, and new outlets are forming, on the one hand, the scale of L2+ level automatic driving mass production continues to soar, and gradually transition to L3 level, on the other hand, the commercial application of advanced automatic driving is also gradually unfolding in specific occasions such as trunk logistics, ports, and mines. Behind the multi-point blossom of autonomous driving, as an important component of the perception layer, the on-board millimeter-wave radar has also stood in the spotlight from behind the scenes and has become a hot spot.

According to data from the Gaogong Intelligent Automobile Research Institute, the millimeter wave radar loading rate of new cars L2+ and above in the domestic market is expected to exceed 50% in 2025. At the same time, 4D millimeter-wave radar will be introduced on a small scale from the beginning of 2022, and it is expected that the number of vehicles will exceed one million by 2023, and the proportion of all forward millimeter-wave radars is expected to exceed 40% by 2025.

Millimeter-wave radar, together with hardware equipment such as lidar and on-board cameras, undertakes the mission of collecting data on the traffic environment around the vehicle, through which the autonomous vehicle can "see" the road conditions clearly, so as to make decisions at any time according to the surrounding environment to ensure safe driving. Compared with lidar and cameras, millimeter-wave radar has all-weather detection capabilities, which can still work normally even under harsh environmental conditions such as rain, snow, dust and fog, and can directly measure the number of distances, speeds, angles, etc., becoming one of the important sensing devices in automatic driving.

However, millimeter-wave radar also has some inherent defects, including the ability to measure "altitude", which makes it difficult to determine whether the stationary object in front of it is on the ground or in the air, and when encountering ground and air objects such as manhole covers, speed bumps, overpasses, traffic signs, etc., it is impossible to accurately measure the height data of the object. In this regard, Cai Qinghan, CTO of Shengke Technology Co., Ltd., pointed out that the 4D millimeter wave radar adds a height analysis of the target on the basis of the original distance, speed and direction data. This makes it possible for 4D millimeter wave radar to make up for the problem of traditional millimeter wave radar, integrating the fourth dimension into traditional millimeter wave radar to better understand and map the environment and make the measured traffic data more accurate.

Shi Lei, executive director of the technical center of the electronic branch of Huayu Automotive System Co., Ltd., summed up the advantages of 4D radar technology, which can be described as three "highs": high detection, ultra-high sensitivity and high resolution. This will make up for many of the problems with traditional millimeter-wave radar.

CICC believes that 4D imaging radar can improve the performance of millimeter-wave radar in all directions, which is expected to make millimeter-wave radar one of the core sensors in the ADAS system and an important direction for the future development of millimeter-wave radar.

Manufacturers accelerate the layout of 4D millimeter wave radar

As attempts to get L3 autonomous passenger vehicles on the road unfold, more and more manufacturers are beginning to work on 4D millimeter-wave radar. Recently, NXP announced that the industry's first dedicated 16nm millimeter wave radar processor S32R45 will be used for the first time in customer volume production in the first half of the year. NXP also said it is developing an L2+-capable millimeter-wave radar processor, the S32R41.

The core devices of vehicle-mounted millimeter-wave radar mainly include single-chip microwave integrated circuits and radar digital signal processors, among which radar digital signal processors are mainly used for digital processing of medium frequency signals of millimeter-wave radar. "Having these processors allows us to support many use cases for millimeter-wave radar, such as near-range environmental mapping, while also further supporting medium-range environmental perception, as well as long-range perception from 300 meters away." With this three-in-one use case, we believe we can drive the large-scale adoption of millimeter wave radar technology in passenger cars above L2+ in the coming years." Steffen Spannagel, global vice president at NXP and general manager of the ADAS product line, said.

Intel's Mobileye is also actively promoting the development and application of 4D millimeter wave radar. Mobileye CEO Amnon Shashua emphasized the use of 4D imaging millimeter-wave radar in automobiles in this year's CES speech, "(By 2025) in addition to the front, we only want millimeter-wave radar, not lidar." ”

In Mobileye's plan, by 2025, it will launch a consumer-grade autonomous vehicle solution based on millimeter wave radar/lidar, equipped with a radar-LiDAR subsystem, when the vehicle only needs to install a forward lidar, plus a 360° all-encompassing millimeter wave radar to achieve autonomous driving tasks.

Or will it replace lidar?

So, is it possible for 4D millimeter-wave radar to replace lidar? Some analysts pointed out that the cost of 4D millimeter-wave radar is similar to that of traditional millimeter-wave radar, and its performance can be comparable to low-beam lidar, which is more in line with the current mass production needs.

The disadvantages of lidar are obvious, that is, fog, rain and snow weather is ineffective, and it is impossible to obtain external images. Cai Qinghan, CTO of Shengke Technology Co., Ltd., said that if the lidar operates in a foggy environment with visibility of only 0.1 kilometers, it has almost no way to detect it. If the lidar operates in a heavy rain environment, the detection distance will be attenuated by about 50%. The measured 4D millimeter wave radar results show that even in the case of rainy weather, the detection range of 4D millimeter wave radar can still reach 300 meters, which is an important reason why 4D millimeter wave radar is more suitable for autonomous vehicle sensors than lidar.

"If you drive the automatic driving mode on the highway, suddenly a heavy rain, the automatic driving function will be invalidated, which is unacceptable." Cai Qinghan said. In contrast, 4D millimeter wave radar inherits the advantages of all-weather anti-jamming of traditional millimeter wave radar, and is not affected by light, smoke, dust, haze, etc., and can work normally at night, rain and snow, etc., and is more adaptable.

Steffen Spannagel believes that "autonomous driving technology cannot rely on a single sensor piece to dominate the world. Based on our understanding of the market, there is no one-size-fits-all sensor, because there are many segments of the market and different levels of autonomous driving, and we think that cameras and radar will coexist because their advantages and disadvantages are very complementary. More special is lidar. We believe there is a great possibility that 4D millimeter wave radar solutions can reduce or replace the use of lidar. 4D millimeter wave radar is still in the early stages of development, but we believe that in the future its performance can be greatly improved and ideally able to eventually replace lidar. ”

Multi-sensor fusion is currently a solution with a relatively high voice. There is a view that multi-sensor fusion is the key to achieving safe redundancy in autonomous driving technology, and 4D imaging radar will obviously occupy a place in it. As for whether it will form an alternative relationship with lidar, it remains to be seen.

Integrated development with deep learning

The current 4D millimeter wave radar is still in the initial cultivation stage before the market starts, and there is still a lot of room for development of product technology. In response to its technology trends, Zhang Hui, CTO of Nanjing Falcon Eye Electronic Technology Co., Ltd., believes that with the development of millimeter-wave radar in the direction of imaging, a clear trend is that traditional pure radar signal processing is developing to deep learning and artificial intelligence processing, which will put forward higher requirements for radar and computing platforms. "At present, on the 4D millimeter wave radar, some companies are already doing related exploration. It is the automatic evolution of the OTA method, or the ability of artificial intelligence to evolve and learn. For example, on Tesla vehicles, based on the ability of video perception, including automatic driving, it can already achieve automatic evolution. In the future, when millimeter-wave radar continues to develop on imaging, its ability to learn and evolve itself must be the future development trend. ”

Shi Lei stressed that multi-sensor fusion is the development direction of millimeter-wave radar and other devices. At present, the development of millimeter-wave radar to 4D radar or 4-piece cascade radar is far from the end of millimeter-wave radar. In the past, the biggest disadvantage of millimeter-wave radar was insufficient resolution, so manufacturers spent a lot of effort and cost to improve its resolution. When the resolution reaches a certain level, continue to increase the resolution through more chip cascading, and its marginal benefits are not so high. A more prominent problem at this point is the lack of dynamic range. Judging from the current 4-chip cascade technology, there is still a lot of room for improvement in dynamic range. If you want to make better and more environmentally aware sensors in the car, there is a clearer and stronger demand for improvement in these aspects.

In addition to being applied to automatic driving, 4D millimeter-wave radar is also rapidly expanding into the field of intelligent transportation, which involves vehicle-road collaboration, that is, the collaboration of "smart cars" and "smart roads", where collaboration is not necessarily obtained only through communication, and may also be coordinated through perception in the future. In response to this, the industry is jointly conducting related explorations.