news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

As national and local laws and regulations on operating autonomous vehicles are successively introduced, the construction of "vehicle-road-cloud integration" is gradually accelerating, and the large-scale implementation of Robotaxi in China has begun to enter a substantive stage.

At the same time, the overseas launch of Tesla Robotaxi was delayed, and there were constant voices from the outside world questioning its technology; but another Robotaxi company, Waymo, has quietly become popular on overseas social networks, forming a sharp contrast between the two.

It is reported that Waymo's Robotaxi adopts a technical route similar to that of most domestic Robotaxi manufacturers such as Luobo Kuaipao, that is, a combination of traditional passenger cars and a large number of sensors, especially lidar.

What is more noteworthy is that Waymo’s current success is closely related to its self-developed Waymo Driver.

In fact, LiDAR will play a pivotal role in both the decision-making weight of autonomous driving and the value of the entire vehicle. Perhaps in the near future, the massive deployment of Robotaxi will greatly affect the commercial prospects of LiDAR.

Perception route competition

For advanced autonomous driving, should we adopt a purely visual approach or rely on lidar?

On this somewhat old-fashioned topic, due to the outstanding performance of Tesla's fully autonomous driving (FSD) and the high cost of lidar, the pure vision route once had the upper hand.

However, recently, there has been a reversal in public opinion on the debate between the pure vision route and lidar.

On July 30, a media outlet released a series of subscription reports and video materials, showing the many accidents that occurred in Tesla in the past few years when the semi-autonomous driving system was turned on, especially a large number of accidents that occurred at night.

The report believes that because Tesla relies too much on the pure visual route and abandons the more expensive lidar due to cost issues, it poses a major safety hazard to passengers.

One thrilling scene was that a Tesla car, while operating on Autopilot, crashed at almost full speed into the chassis of a truck that had overturned and was lying across the middle of the road at night, causing casualties among the passengers in the car.

Last year, another media outlet also reported on Tesla's autopilot accidents, and said that the actual number of Tesla accidents was far higher than the public number.

If the use of lidar in private cars is an "option", then the use of lidar in Robotaxi may be standard.

"Cameras are passive sensors. From the perspective of physical principles, in low light or no light, it is pitch black. If the color of the obstacle is the same as the background, the camera will misjudge it." A LiDAR industry insider told the 21st Century Business Herald reporter, "LiDAR can accurately solve these scenarios, so all Robotaxis are equipped with LiDAR."

Another autonomous driving algorithm practitioner told the reporter that in addition to the problems of line of sight and vision, the small number of safety accident cases is also the reason why the pure vision route cannot completely replace the core of lidar for the time being.

"When a vehicle is equipped with L4 autonomous driving functions, the requirements for safety redundancy are higher; when the pure vision solution based on deep learning algorithms has not yet achieved full road condition coverage, its safety is still questionable, and the multi-sensor fusion solution is more reliable, so the existing Robotaxi solutions all adopt this route." Founder Securities said in its research report, "The vision solution is the development trend of the industry, but radar needs to be supplemented for safety redundancy. The number of radars installed is expected to decrease, but it is still an indispensable hardware for achieving safety perception."

The number of LiDARs installed has increased dramatically

At this stage, Robotaxi's demand for lidar is almost dependent.

Hesai Technology, a leading domestic LiDAR company, told the 21st Century Business Herald reporter: "LiDAR is an essential sensor for L4. Robotaxi mainly relies on LiDAR and high-precision maps for mapping and positioning.

According to public information, all Robotaxi vehicles in the testing and operation stages are equipped with LiDARs, without exception. Some Robotaxi vehicles are equipped with up to 8 LiDARs.

According to statistics from Founder Securities, the number of LiDARs installed in Baidu's Apollo RT6 solution is as high as 8. In addition, Pony.ai's sixth generation, WeRide Sensor Suite 5.0, and Didi's Gemini solution are equipped with 7 LiDARs respectively.

The two major overseas Robotaxi giants, Waymo and Cruise, owned by General Motors, are both equipped with five laser radars.

In comparison, the number of laser radars equipped in C-end smart cars is far less than that in Robotaxi.

"In the past, Robotaxi mainly used 360° fully rotating mechanical radars combined with a variety of blind spot radar product combinations. Most C-end smart electric vehicles are equipped with 1 to 2, such as the Ideal L series, Xiaomi SU7, etc. There are also models like BYD Yangwang, Avita 11, 12, etc., which are equipped with 3 laser radars per vehicle." Hesai Technology told reporters.

"Both in terms of product performance and the number of LiDARs, Robotaxi has higher requirements, because L4 autonomous driving requires stronger and more comprehensive perception capabilities, as well as higher safety redundancy, in order to fully realize the replacement of human drivers with machines."

An expert from RoboSense also told the 21st Century Business Herald reporter: "Globally, LiDAR is a must-have sensor for Robotaxi. The core reason why Robotaxi is equipped with LiDAR is safety. Robotaxi is L4+ autonomous driving. The vehicle operator needs to be responsible for safety. Only by equipping LiDAR can the safety be high enough to ensure the safety of intelligent driving."

Robotaxi sales increase

Some industry experts pointed out that the Robotaxi previously used for road testing or trial operations had a single vehicle purchase cost of 200,000 yuan, while the optical sensor also cost 200,000 yuan, and the cost of algorithms and computing equipment was around 100,000 yuan.

At present, the cost of a single vehicle and sensor has dropped to around 100,000 yuan, of which LiDAR accounts for a relatively high proportion. As the cost of the entire Robotaxi host is halved, it can effectively stimulate the demand for LiDAR through fleet construction and replacement of old models.

"In fact, many of the current cylindrical mechanical laser radars mounted on the top of Robotaxi are non-automotive-grade products. Their lifespan and reliability are not as good as the automotive-grade laser radar products used in C-end passenger cars. Of course, Robotaxi is also gradually upgrading to automotive-grade laser radars." The aforementioned RoboSense expert told the 21st Century Business Herald reporter.

"In terms of performance, Robotaxi has higher requirements for LiDAR. For example, it needs 360° horizontal coverage, which can be provided by one mechanical LiDAR. Four vehicle-mounted LiDARs are generally required, but the cost of the latter is still lower than that of the former, which is the advantage of mass-produced products for vehicle specifications." Hesai also told reporters that the earliest Pandar provided by the company for Robotaxi cost about 200,000 yuan per unit. At present, the LiDAR of ADAS products only costs 500 US dollars per unit, and even a combination of four costs only 2,000 US dollars. The future decline in sensor prices will greatly stimulate the decline in Robotaxi costs, making it even lower than the price of some C-end electric vehicles.

It is reported that Hesai and RoboSense, as two strong players in the field of lidar in China and even the world, also occupy the mainstream of the current Robotaxi sensor market.

Among them, as of 2023, Hesai has supplied lidar to well-known Robotaxi companies including LoBo, Pony.ai, WeRide, Cruise, Aurora, Didi, AutoX, and Zoox.

Experts from RoboSense revealed that the company also provides RoboSense's M-series automotive-grade lidar to L4 autonomous driving operating vehicles under LoBo, Pony.ai, and WeRide.