Accounting for as much as 1/3 of the cost, is the laser radar used in driverless cars a "big scam"?

With the iPad and iPhone both using lidar, the discussion on the use of automotive lidar has become increasingly heated. After all, who can refuse to own a car that can drive on the road by itself and safely?

Some people may say that Tesla and other manufacturers already have smart driving cars, but judging from the current so-called intelligent effects, they are still far from the level achieved by driverless cars after the application of lidar. Therefore, some people say that only cars that use lidar are truly driverless cars.

1. What is LiDAR?

LiDAR (Light Detection And Ranging) is a product of the combination of laser technology, electronic technology and signal processing technology. A standard LiDAR system contains an excitation source, a laser, an optical detector, a signal conditioning circuit, and a computer control and information processing system.

Rotating mirror laser radar

Source: Sensor Expert Network

Velodyne's 64-line LiDAR

Source: Leifeng.com

From the first time lidar was used in driverless cars to the present, different forms of vehicle-mounted lidar have been developed. From a mechanical structure point of view, they can be divided into three types: mechanical, hybrid solid-state, and pure solid-state.

The cylindrical laser radar commonly installed on the roof of a driverless car is a mechanical laser radar, whose internal laser can rotate 360 ​​degrees to comprehensively scan the surrounding environment. Currently, mass-produced models with advanced driver assistance functions are equipped with planar laser radars on their bodies, most of which are hybrid solid-state laser radars.

LiDAR installed on the roof

Source: Baidu

LiDAR scans the surrounding environment

Source: Vico.com

The working principle of lidar is that the excitation source acts on the laser, which then emits a laser beam at the target object. The reflected signal is then received by the photodetector. The received signal is sent to the computer control and information processing system for comprehensive information processing after passing through the signal conditioning circuit.

Simply put, it is to send out lasers first and then receive them, and then process the signals obtained, and finally determine how far away the target object is, where it is, how high it is, how fast it is, what it looks like, and the preliminary judgment of its material, etc. In a sense, LiDAR is like the sonar system of bats, with the functions of measurement and scanning.

Source: Qilu Laser Sharing Industrial Park

In autonomous driving, lidar will continuously emit and receive lasers to obtain information about the surrounding environment, and send the information to a computer, which will then be integrated with information from other sensors so that the system can make correct decisions, including braking, obstacle avoidance, choosing the optimal route, and so on.

Source: CSDN

2. What technology does a driverless car use?

Most smart cars on the market with advanced driver assistance functions choose to use millimeter-wave radar, ultrasonic radar and cameras to realize the perception and recognition of surrounding objects. Taking Tesla as an example, Tesla currently mainly uses cameras to detect the surrounding environment, combined with ultrasonic radar and millimeter-wave radar installed around the car body, to see and hear everything.

Using mmWave radar to monitor gestures

Source: Hackaday

Tesla has installed a large number of cameras and sensors on the body of its car. Although "three cobblers are better than one Zhuge Liang", accidents involving Tesla's driverless cars still occur from time to time. This is mainly because Tesla mainly relies on visual perception and learns various scenes on the road in advance. It cannot recognize special scenes that it has not learned. Other types of radar sensors also have problems with accuracy and scene applicability.

Tesla electric car sensing area and range

Source: Tesla

If lidar is used, the system can identify obstacles and collect information more accurately, and the chip installed in the car can make more accurate instructions based on the rich information.

Analyzing the autonomous driving technology routes of various relevant manufacturers at home and abroad, IT companies such as Google and Baidu aim to directly achieve fully autonomous driving; car companies such as BMW, GM, Ford, Volvo and Tesla hope to achieve the transition from assisted driving to fully intelligent cars in a progressive manner, and they have different attitudes on whether to adopt lidar solutions.

3. It is difficult for LiDAR to enter the autonomous driving market

Since LiDAR has so many advantages, why don't self-driving car manufacturers choose to use this technology? The biggest reason is that it is expensive.

In the early years, except for Tesla, almost all mature autonomous driving technology solutions used 64-bit laser radar, which costs about RMB 700,000. A small radar is equivalent to a car or even several cars. For most consumers, the autonomous driving function should not occupy most of the cost of the car. For a car worth RMB 300,000, paying RMB 100,000 for the autonomous driving sounds acceptable; for a car worth RMB 1 million, paying RMB 700,000 or 800,000 for the autonomous driving is a bit like buying a box and returning the pearl.

LiDAR price of representative manufacturers

Recently, as more and more manufacturers enter the market, the price of LiDAR is gradually decreasing, but the price of nearly $1,000 per unit is still too expensive compared to other sensors on the vehicle body. Moreover, to achieve a higher level of autonomous driving, a vehicle needs more than one LiDAR. For example, the Alpha S Huawei HI version uses 3 LiDARs, 13 cameras, 6 millimeter-wave radars and 12 ultrasonic radars.

LiDAR installation status of representative models

Source｜See watermark

Million-dollar models like the Audi A8 have only reached the L3 level, which means conditional autonomous driving, not fully mature autonomous driving that does not require any human control.

Only by continuously reducing the cost of LiDAR can it be mass-produced. Even if the cost problem is solved in the future, manufacturers still have to adapt the software to LiDAR. Guidehouse, a public and commercial consulting company, conducted a comprehensive evaluation of 15 global autonomous driving companies based on technology, R&D progress, product portfolio and other dimensions. Guidehouse divides these companies into four levels: "leader", "competitor", "challenger" and "follower". Baidu is the only Chinese company on the list of leaders.

The latest autonomous driving competitiveness list

Source: Guidehouse

If LiDAR is to be popularized, China, as a country that encourages technological innovations such as autonomous driving, will play an important role. This requires close cooperation among domestic high-tech autonomous driving companies, vehicle manufacturers, IT companies, and LiDAR research and development companies to focus on breaking through technical difficulties and practical application problems, including how to improve measurement accuracy and range, how to improve angular resolution, how to enhance point cloud density and stability, and how to improve radar's ability to resist environmental interference, etc.

LiDAR is the brightest eye that can be installed on driverless cars at present. As the technology becomes more mature, the shining eyes will inevitably be installed on every household's car. It is not difficult to imagine that in the near future, civilian LiDAR will be promoted and truly driverless cars will enter thousands of households.