Ultrasonic, Lidar, Millimeter Wave Radar in Autonomous Driving

Currently, there are three main types of radars in unmanned vehicles: ultrasonic radar, lidar, and millimeter-wave radar. Although the latter two radar technologies are rising stars, their role in autonomous driving has become increasingly prominent in recent years.

Every year, 1.25 million people die in car accidents. Human error (drunk driving, speeding, ignoring traffic lights, texting while driving) caused more than 94% of fatal accidents. To reduce automobile accidents as close to zero, automobile manufacturers, automobile suppliers, governments, and academia providers are jointly developing Self-driving cars. The establishment of a new automotive ecosystem of autonomous vehicles requires the support of multiple advanced technologies such as sensor fusion, a new automotive network architecture, and the Internet of Vehicles.

Ultrasonic radar: mature technology and low cost

We can see ultrasonic radars on the front and rear sides of many vehicles. According to the working frequency, there are three types of ultrasonic radar: 40kHz, 48kHz, and 58kHz. The higher the frequency, the higher the sensitivity and the smaller the detection angle.  The ultrasonic radar can measure obstacles within the range of 0.2-5m with the accuracy of 1-3cm by sending and receiving ultrasonic waves.

However, because the operating frequency belongs to the acoustic wave range, the shortcomings of the ultrasonic radar are also obvious. Especially when a car is driving at a high speed, the received information will be delayed due to the propagation delay of the ultrasonic signal. In addition, ultrasonic equipment also has the problem of poor directivity, requiring more equipment to cover the same area. And weather conditions will greatly affect their detection effects.

However, this has not affected the application of ultrasonic radar in the automotive industry. A very critical point is its super high-cost performance. The price of a single ultrasonic radar on the market is only tens of yuan. According to a set of reversing radar systems to install 4 ultrasonic radars, the hardware cost is less than 200 yuan. We know the automatic parking system requires a larger number of ultrasonic radars. However, the total hardware cost can also be controlled at about 100$. Compared with lidar that costs tens of thousands of dollars at every turn, the cost advantage of ultrasonic radar is too prominent. Tesla has a special preference for ultrasonic radar, using 8 in parking assist and 12 in assisted driving.

Lidar: indispensable for autonomous driving

LiDAR (light detection and ranging), which means laser detection and ranging, is a sensing technology that emerged after radar and sonar. It uses laser pulses to scan the environment instead of radio waves or sound waves. The wavelength is nanoscale. Among all the sensing technologies used in automobiles, lidar is a novelty, which plays an important role in autonomous vehicles. Lidar can provide the most accurate three-dimensional map and can scan the 360-degree space around the self-driving car, with a range of up to 100 meters. Some lidar systems even provide up to 64 channels and scan more than 1 million points per second. This amount of information can provide high accuracy of 2 cm to cope with the ever-changing environment. In addition, the reflectivity of the laser signal can also distinguish the material of the target substance.

Lidar

Airborne lidar is mainly on helicopters or drones to collect data. Ground-based lidar systems are usually installed on moving vehicles or tripods on the ground to collect accurate data points. Self-driving cars use a mobile ground lidar system. Only one lidar can scan multiple angles at the speed of light at the same time to create a detailed three-dimensional image or map of the surrounding area. Lidar uses lasers and mirrors for echo imaging and maps the surrounding environment in real-time. Therefore, its measurement accuracy is significantly higher than that of ultrasonic radar. Today, we can find lidar in many key automotive and mobile applications, including advanced driver assistance systems (ADAS) and autonomous driving systems. This technology combines with other sensory data to provide a more reliable representation of static and moving objects in the vehicle environment.

Millimeter-wave radar: moving towards 77-79GHz

Compared with the complicated mirrors and lasers of lidar, millimeter-wave radar is much simpler. But its signal transmission speed is faster and more accurate than ultrasound. Millimeter-wave radar usually works at 24GHz and 77-79GHz. Basically, it does not afraid of environmental factors such as heat or light. Millimeter-wave antennas are also much smaller than ultrasonic antennas and have lower power. We can adjust them to short, long, wide, or narrow detection ranges to meet the needs of specific applications. The traditional 24 GHz narrowband automotive radar has certain limitations in distinguishing objects and distinguishing people, dogs, and other cars. At present, automotive radar sensing technology with 24GHz narrowband sensors as the mainstream develop rapidly in the direction of the 76-81GHz frequency band, frequency modulated continuous wave (FMCW), and beamforming antennas.

With the transition from 24GHz to 79GHz, the performance in terms of range resolution and accuracy can be improved by nearly 20 times. In other words, the distance resolution of the 24GHz system is 75cm. If it is a 79GHz system, the resolution will reach 4cm, which can better detect multiple adjacent objects. Similarly, in the case of smaller wavelengths, the resolution and accuracy of speed measurement will also be proportionally improved. Another advantage of is that the size and weight of the equipment will be significantly reduced.  The wavelength of the 79GHz signal is about one-third of that of the 24GHz system. Therefore, the total area of ​​the 79GHz antenna is only one-ninth of that of the 24GHz antenna. Developers can use smaller and lighter sensors and easily hide them for better fuel economy and car shape design.

Fusion of automotive sensor technology

In addition to the above three sensing technologies, vehicle-mounted cameras are also sensors in automobiles. Due to space limitations, this article does not specifically introduce them. It is undeniable that in current vehicles, the on-board camera serves as the main visual sensor of the ADAS system. The lens collects the image, which is then processed by the camera’s photosensitive and control components. Then it’s converted into a digital signal that can be further processed by the computer to realize the perception of road conditions around the vehicle, including forward collision warning and lane deviation Alarm, pedestrian detection, and other functions.

Any kind of sensing technology has its advantages and limitations. The automobile industry will not rely solely on one sensing technology to realize its autonomous driving function. Most manufacturers combine these three sensing technologies. The purpose is to ensure that their autonomous driving systems obtain reliable data in terms of range, resolution, and robustness.

The current sensing solutions still cannot provide enough data to realize fully automatic driving. There are still many problems in the current autonomous driving technology. However, they have greatly reduced the driver’s human error through the ADAS system. In turn, the adoption of ADAS technology and driverless cars have promoted the development of the automotive sensor. According to market research firm MarketsandMarkets, by 2025, the automotive sensor market will grow from $24.5 billion to $40.3 billion. The compound annual growth rate of 10.5% from 2020 to 2025. The global production of motor vehicles and the increase in demand for automotive electrification, including consumer demand for automotive safety and comfort, are the main factors driving the growth of the automotive sensor market.