Five vehicle sensors (millimeter wave/laser/ultrasonic radar/infrared sensor/camera) Hengping

"This road is really blocked, it would be great if the car can drive by itself!"

I believe that almost all drivers have thought about this problem when they encounter serious traffic jams. If the car can drive automatically, That car will become a good place to recharge and recharge, and this is also one of the typical scenarios where many car factories promote intelligent driving. Good things are afraid of turning words like "but" and "but". For fully autonomous driving, the vision is good, but the status quo is that the car factory shouts, "Concubines can't do it." The biggest problem that cannot be achieved is that the on-board sensors do not meet the requirements.

Many people may not have an impression of the 2016 Tesla crash. In January 2016, Mr. Gao, a 23-year-old Tesla owner, drove his Tesla on the Handan section of the Beijing-Hong Kong-Macao Expressway in Hebei Province. He unfortunately died because he could not avoid a rear-end collision with a road sweeper in time. One year after the accident, Tesla admitted that the car was in an autonomous state. Although Tesla’s autopilot has been upgraded in recent years, accidents have occurred occasionally, hitting trailers, cement piers, fire trucks...

You know, even the first generation of Autopolit has multiple Cameras, millimeter wave radars and ultrasonic radars do not guarantee 100% zero accidents. For Tesla, which emphasizes "fully autonomous driving", this is more like an over-marketing.

In addition to the above three sensors, infrared sensors and lidar are also associated with automatic driving. The development speed and height of the five major sensors determine the speed of the future progress of automatic driving. From the current situation, it can only be said that the five sensors have their own advantages and disadvantages.

millimeter-wave radar

It can be said with certainty that millimeter-wave radar must have a place in the future true self-driving cars, because it has all-weather capabilities. Whether it is fog, smoke, rain, snow, sunlight, etc., millimeter-wave radar can respond calmly.

In addition to all-weather and all-weather, the millimeter-wave radar has another four sensors that cannot be matched by the long test distance, which can reach more than 1000 meters.

According to the frequency of the current mainstream vehicle millimeter wave radar, there are two main types: 24GHz mm Wave radar and 77GHz millimeter wave radar (79GHz in Europe). Generally, the detection range of 24GHz radar is medium and short distance, used to realize BSD (BlindSpotDectection, blind spot detection system), and 77GHz long-range radar is used to realize ACC (Adaptive Cruise Control, adaptive cruise system). The emergence of

77GHz also makes millimeter-wave radar take the lead in device size. Due to the shorter wavelength, the antenna size on the device is drastically reduced. However, more advanced packaging methods are required and the cost is higher. The disadvantage of

millimeter wave radar is that its static ranging is too complicated, and the judgment of tangential motion is poor. From the cost point of view, the price of millimeter wave radar is second only to lidar and is a more expensive device.

From an industrial perspective, the initiative of millimeter-wave radar is still in the hands of international giants. Traditional component giants such as Bosch, Continental, Autoliv, and Delphi have basically completed the monopoly from R&D to application. The four mentioned are the well-known "ABCD" in the automotive industry. In addition to a few of them, the shipments of Denso and Hella also make domestic companies look ahead.

For domestic millimeter wave radar manufacturers, both 24GHz and 77GHz are involved. Among them, the development progress of Xingyidao is worth mentioning. Since its establishment in 2014, Xingyidao completed the 77GHz sample in the second year and completed the commercialization of 77GHz in the third year. It is China's first and OEM cooperationMeadow radar enterprise. The difficulty of

77GHz millimeter-wave radar lies in the high process specifications and high RF chip requirements, which can overcome these two problems, which shows the real skill of easy roads.

In addition to easy roads, domestic millimeter wave radar manufacturers also include Zhibo Technology, Senstech, Trotec, Falcon Technology, etc., and they also have 24GHz and 77GHz layouts, but in terms of accuracy and distance The product gap with international manufacturers is still obvious.

Ultrasonic radar

Among vehicle sensors, ultrasonic radar is one of the most common types at present. In short-distance measurement, ultrasonic ranging sensors have very big advantages. It is mostly used in reversing radar. The principle of the

ultrasonic radar is very simple, that is, the ultrasonic wave is sent out through the ultrasonic transmitting device, and then the distance is measured by the time difference when the receiver receives the transmitted ultrasonic wave.

There are two common ultrasonic radars. The first is the reversing radar installed on the front and rear bumpers of the car, which is used to measure obstacles in the front and rear of the car. This type of radar is called UPA in the industry; the second is installed on the side of the car to measure side obstacles. Ultrasonic radar for object distance is called APA in the industry.

Currently, the working frequencies of commonly used probes are 40kHz, 48kHz and 58kHz. Generally speaking, the higher the frequency, the higher the sensitivity, but the smaller the horizontal and vertical detection angle, so 40kHz probes are generally used.

Compared with the other four sensors, the biggest advantage of ultrasonic radar is that it is cheap. Of course, it's not that the ultrasonic radar is cheap and the performance is bad. Ultrasonic radar has the best angle measurement capability, while data processing is simple, and it has good penetrability. The disadvantage of

is that the range of ultrasonic radar is very short, the maximum distance is 15 meters, and the specifications are usually several meters. Ultrasonic radar is very poor for tangential measurement and is easily interfered by weather factors when used.

The technical solutions of ultrasonic radar generally include analog, four-line digital, two-line digital, and three-line active digital. The latter three increase in signal interference processing effects. Of course, the price and equipment difficulty are also increasing. .

From an industrial point of view, Bosch and Valeo are giants in the industry, because these two control the main markets for reversing ultrasonic radars. In addition to these two, international manufacturers such as Japan’s Murata, Nisera, and Electric Zhuang, Mitsubishi, Panasonic, etc. are also very influential. Domestic manufacturers include Taiwan Tongzhi Electronics, Shenzhen Hangsheng Electronics, Shenzhen Horn, Hui Chuang, Shang Fu, Audi Wei, etc. There are many manufacturers, but few can enter the automotive pre-installation market. Tongzhi Electronics is one of the rare ones. Tongzhi Electronics mainly produces products such as car reversing radar, remote control, rear-view cameras, smart interior mirrors, etc. It is a major domestic automobile manufacturer (such as Shanghai GM, Shanghai Volkswagen, Dongfeng Nissan, Shanghai Automobile, Shenlong Automobile, Chery Automobile, Geely Automobile, Ford Motor, etc.) suppliers, and is currently the number one supplier in the Asian parking sensor OEM market.

LiDAR

LiDAR is abbreviated as LiDAR. It is a system that integrates laser, GPS global positioning and inertial measurement devices to obtain data and generate accurate digital models. Lidar is a relatively new field, but its popularity is rising rapidly, and it has taken a place in the field of autonomous driving. According to statistics, the Chinese market alone will reach a market size of 600 million yuan by 2021.

Among the five sensors, the range of lidar is second only to millimeter wave radar, and the maximum range exceeds 300 meters. At the same time, lidar can generate a three-dimensional position model, so it has a good angle measurement capability. In addition to the above, the advantages of lidar include higher measurement accuracy, while the response speed is more sensitive, and it is not affected by ambient light.

Of course, lidar also has its own shortcomings. The first thing to mention is the high price. Although the solid-state lidar has effectively controlled the cost after its appearance, the price of lidar is still expensive compared to other sensors.

Lidar can also be divided by wire beam, which is divided into single-beam and multi-beam lidar. Single-beam radar scans only generate one scan line at a time, so it generates plane information, but the measurement speed is fast, and it is used for topographic surveying and other aspects; multi-beam lidar has a larger scanning angle and more accurate data, so currently Lidar products on the market include 4 wire harness, 8 wire harness, 16 wire harness, 32 wire harness, 64 wire harness and 128 wire harness.

In the automotive field, lidar can be used in many aspects such as automatic parking, ACC active cruise, automatic braking, and unmanned self-driving.

At present, there are many LiDAR manufacturers at home and abroad, but the number of applications in the field of autonomous driving is limited. The main reason is that the price is high. It is difficult to see LiDAR cars with less than 10,000 yuan on the market. Module.

Representative foreign manufacturers are represented by Velodyne and Quanergy.

In recent years, Velodyne's 3D lidar business has developed rapidly. The LiDAR sensor used by Google in its earliest autonomous driving prototype car was developed by the company. The measurement accuracy of its products is at the benchmark in the entire industry. Position and performance are superior. The price of Velodyne's autonomous driving lidar products is between 38,000 and 700,000 yuan.

Quanergy was established in 2012. In September 2014, Quanergy and Mercedes-Benz reached a strategic cooperation to develop in-vehicle sensing systems and unmanned vehicles for Mercedes-Benz. In October 2015, Quanergy announced a collaboration with Delphi to develop a new type of lidar system for driverless cars. Currently, Quanergy has launched two 3D lidar products, one is TheMarkVIII installed on the Mercedes-Benz smart driving test car, and the other is the first all-solid-state lidar S3 designed for smart driving cars.

There are also manufacturers of automotive lidar in China, and Sagitar Juchuang, Beixing Photonics, and Beike Tianhui have launched related products.

Car camera

What the car of the future needs to achieve is to "hear six directions, see all directions", and the "eye" of the car is these car cameras.

When the autonomous driving concept has not yet heated up, the car has begun to appear on the camera device, the function is to realize the driving record As well as reversing images, etc., as ADAS and autonomous driving become more and more popular, the functions of on-board cameras have begun to become more abundant. The surge in the number of assemblies has also made in-vehicle cameras a new blue ocean in the industry. To realize ADAS, six cameras are needed, and Tesla's L3 level of autonomous driving uses eight cameras. Market research organization HIS predicts that in 2020, the shipment of car cameras will reach 87 million. Therefore, whether it is the pre-installation market or the post-installation market, the market size of the car camera will not be small.

is so optimistic that in-vehicle cameras do have many unique advantages in the field of autonomous driving. It is the only sensor that can clearly distinguish and identify road targets. It can guarantee the collection of stable data in complex sports road conditions, and the pixel requirements are not high. 300,000-1.2 million pixels can be satisfied, so the price There are great advantages in this regard.

Of course, the vehicle-mounted camera also has its own disadvantages, of which light interference is the biggest problem, and it has no ability to accurately control the speed and distance. It must cooperate with other sensors.

At present, whether the car camera is in the mainstream CMOS market or the whole market, the giant effect is very obvious. Sony, OmniVision and Samsung occupy more than 70% of the market. If you talk about the CMOS market alone, Aptina and OmniVision are the companies with the highest market share.

In addition to the Chinese-funded OmniVision, there are also domestic CMOS design companies such as Gekewei and Spyco, which are rising rapidly.

Infrared sensor

At present, if you want to comment on the reputation of these five sensors, then the infrared sensor is indeed the least famous. We have seen many sensor combinations for autonomous driving, including millimeter wave radar + ultrasonic radar + camera, and laser radar + ultrasonic radar + camera. There is really little mention of infrared cameras. However, after Uber's self-driving vehicle had a high-profile fatal accident in Arizona, the industry did have to re-examine whether to install infrared sensors to build safe self-driving cars.

The current status of the industry is that the camera's role in low illumination and light conditions is greatly reduced, and the millimeter wave radar is used to identify obstacles. The ability is still zero, and ultrasonic radar is beyond its reach. Lidar does have this ability, but the weather has too much influence on Lidar. At present, research institutions have confirmed that based on infrared night vision systems that use infrared light waves to detect differences in heat emitted by objects naturally, objects that cannot be identified by visible light cameras, radar, and lidar can be detected.

In terms of performance alone, the advantage of infrared sensors is that they can distinguish tangential motion well and can more accurately identify organisms. In addition, another advantage is the low price. However, there are many shortcomings of infrared sensors. The first is that the ability to discriminate radial motion is very poor, without the ability to measure angles, it is impossible to complete static ranging.

Judging from the current form, the future of fully autonomous driving cannot do without the help of infrared sensors. Infrared sensors are the only technology that can classify, identify and explore the surrounding environment of vehicles in any environment or weather conditions. Autonomous vehicle developers predict that future vehicles will be equipped with several far-infrared cameras to expand their detection range and better provide images of the surrounding environment of the vehicle.

Of course, the current in-vehicle infrared sensors are only recognized, and few companies officially launch products in the development stage.

Summary

In addition to traditional car factories, companies that develop autonomous vehicles also have many technology companies, and of course there are also startups. Each company has a different size and a different position. The understanding of driving is also differentiated, and the product mix and the number of products used vary greatly. However, no matter how to integrate these sensors, driving safety is a common goal. The fatal car accidents of Tesla and Uber have given us a warning. It is irresponsible to open tests or even publicly use autonomous driving aggressively. The hard work of the hardware is not in place, and even the smarter algorithm will be "inattentive".