L3 autonomous driving standards are expected to be released. Industry insiders: It can drive the rapid development of related industries

According to Xinhua News Agency, Xin Guobin, Vice Minister of the Ministry of Industry and Information Technology, stated at a regular policy briefing held by the State Council Information Office that he would launch a pilot program for the admission and on-road access of intelligent connected vehicles, and organize the launch of city-level "car-road cloud services". "Integrated" demonstration application to support conditional autonomous driving.

"This is about the commercial application of L3 and higher autonomous driving functions. We have already started this work, and local governments are also responding actively. I believe that in the near future, more intelligent products will run On the way." Xin Guobin said.

Source: SenseTime official website

In addition, the Ministry of Industry and Information Technology revealed in mid-May that the "Intelligent Connected Vehicle Standard System Guide" will be officially released soon, accelerating the formulation of more than a dozen key and urgently needed standards, including supporting the commercial application of L3 and above autonomous driving functions.

Bai Yiyang, vice president of the Research Department of CMB International, said that the formulation of intelligent connected vehicle standards can clarify aspects such as safety standards and division of responsibilities, and guide the development direction of the industry. Regarding the development of L3, we mainly look at two aspects. On the one hand, autonomous driving is an industry development trend and can well drive the development of related industries. On the other hand, my country has a certain first-mover advantage in the layout of autonomous driving. In the future, it will have the opportunity to become another area leading the development of the automotive industry after electrification.

According to Gaogong Intelligent Automobile Data, from January to March 2023, the penetration rate of L2 level and above in the passenger car industry was 33.4%, a year-on-year increase of 9 percentage points; the penetration rate of L2 level and above of new power brands is basically above 70%, which is the first The penetration rate of some self-owned brands at L2 level and above is over 30%. It is not difficult to see from the data that the jump from L2 to L3 in domestic autonomous driving has become an irreversible trend.

Hong Yong, an associate researcher at the E-Commerce Institute of the Ministry of Commerce Research Institute, said that countries around the world are actively promoting the commercialization of autonomous driving. However, due to the large differences in road environments, regulations and technical standards between countries, the commercialization of autonomous driving may have delays. different. Nonetheless, with technological advancement and policy support, it is expected that more self-driving services will be put into use around the world in the next few years.

However, due to the characteristics of L3 autonomous driving, there are still some difficulties in promoting its development.

Frequent accidents make it difficult to define responsibility

L2-level or even L2+-level automatic driving assistance functions provide users with driving convenience, but sometimes they also become safety hazards.

On August 12, 2021, Lin Wenqin, the well-known catering founder, had a traffic accident on the Hanjiang section of the Shenhai Expressway after driving  a Weilai ES8  (  parameter  |  inquiry  ) car with the automatic driving function (NOP pilot status) enabled. He died in 2021. Only 31 years old.

On June 17, 2022, when a Xiaopeng P7 was driving on the highway, the LCC lane centering assist driving function was turned on. At about 120km/h, the vehicle suddenly automatically changed lanes to the right, almost causing an accident.

On March 13, 2022, Mr. Deng from Yueyang, Hunan Province was driving a Xiaopeng P7 on a national highway. After turning on the automatic driving system and driving for more than ten kilometers, he suddenly encountered a car that rolled over on the road. However, his car did not have any alarms or warnings. He slowed down and hit the car directly. After discovering the abnormality, he applied the brakes urgently, but the car did not respond.

The California Highway Patrol said on July 18, local time, that the U.S. National Highway Traffic Safety Administration (NHTSA) was launching a "special investigation" into a recent fatal accident in South Lake Tahoe, California. The vehicle involved was a 2018 Tesla Model 3, which collided head-on with a 2013 Subaru Impreza. At that time, the Tesla car was suspected of relying on its automatic driving assistance system to drive.

Source: SenseTime official website

In February this year, a fatal accident also occurred in Contra Costa County, California. A 2014 Tesla Model S crashed into a fire truck of the California Fire Department. The Tesla driver died on the spot. In March this year, a 2022 Tesla Model Y hit a 17-year-old student who got off the car in North Carolina, causing serious injuries to the latter.

Reuters reported that since 2016, U.S. auto safety regulators have investigated Tesla for more than 30 special accidents on suspicion of using autonomous driving and other systems. As of July 18, local time, 22 accidents resulting in the death of drivers or passengers have been reported.

The difficulty in defining liability after an accident has become the primary problem for the commercialization of L3 autonomous driving.

At the 2022 World Artificial Intelligence Conference (WAIC), Robin Li judged that after L2, it is likely that L4 will be the first to enter commercialization, not L3.

The reason given is that the liability for accidents at L2 and L4 is clearly defined. The L2 responsibility lies with the driver, and the L4 operator is responsible for the accident. L3 is different, the driver takes over when needed, which makes it difficult to determine responsibility for the accident.

Autonomous driving is usually divided into six levels, from L0, L1 to the highest L5. Different technical levels correspond to different responsibilities. Among them, the L0-L2 stage is considered to be assisted driving, where the driver is the leader, while L4 and L5 are machine-led unmanned driving stages with relatively clear responsibilities.

Although the L3 level autonomous driving system, which is in the transitional stage, is considered a "watershed" in self-driving capabilities, marking the official transfer of driving rights from "humans" to "autonomous driving systems", it is precisely this positioning that puts L3 in a It’s an awkward position. Since both the driver and the machine need to be involved, who is responsible?

Car companies have given up on L3

In 2017, Audi released the world's first new sedan "Audi A8" that can achieve L3 level autonomous driving. This model is also defined as the first mass-produced model developed specifically for L3 level autonomous driving.

Subsequently, some domestic car companies began to follow suit. Geely, SAIC, Changan and GAC have all revealed plans for mass production of L3 autonomous driving models. SAIC Roewe MARVEL X Pro and GAC Aion LX released in 2019 already have L3 autonomous driving functions. In 2020, Changan Automobile launched UNI-T equipped with the country's first mass-produced structural road L3 autonomous driving system.

However, by 2020, after spending 9 years, investing a thousand-person team and 1 billion euros in research and development expenses, Audi announced that it would abandon the world's first mass production plan for L3 autonomous driving.

Many foreign car companies such as Volvo and Ford have stated that they will go beyond L3 and directly develop L4/L5 level autonomous driving technology. The main reason is that there are currently no laws, regulations and testing standards applicable to L3 autonomous driving worldwide.

Volvo CEO Hakan Samuelsson believes that L3 is unsafe. Volvo is currently cooperating with Uber to develop L4 level autonomous driving, and has rolled out the XC90 with L4 autonomous driving hardware for Uber to test.

Domestically, Li Auto has stated that it has skipped the research and development of L3 autonomous driving and is focusing on L4 Robotaxi.

Currently, only Mercedes-Benz has truly realized the implementation of L3 autonomous driving. Mercedes-Benz has officially announced that its autonomous driving assistance system Drive Pilot will be available for ordering in Germany on May 17. It will be available as an option on S-Class and EQS models. The vehicle will have L3 level autonomous driving capabilities.

L3 still has a long way to go

Xu Haidong, deputy chief engineer of the China Association of Automobile Manufacturers, once said that China cannot tie its own hands when it comes to the market-based application of high-level autonomous driving and must liberalize in a timely manner to accelerate its marketization process.

However, there is still a long way to go before L3 level autonomous driving can achieve large-scale commercial mass production.

On July 17, at the 5th World Intelligent Safety Conference hosted by FISITA (World Federation of Societies of Automotive Engineers) and the China Society of Automotive Engineers, Li Jun, academician of the Chinese Academy of Engineering and chairman of the China Society of Automotive Engineers, said, "The next three years will It is a critical period for the commercialization of high-level intelligent connected vehicles.” Countries around the world have begun to issue international certifications for L3 level autonomous driving systems. This year, China’s Ministry of Industry and Information Technology and the Ministry of Public Security took an important step in the field of autonomous driving, clearly expressing their support for the commercial application of L3 and higher-level autonomous driving functions.

However, Li Jun said that currently, autonomous driving faces challenges such as road traffic safety, the industry's lack of complete system technical standards, functional failures, cyber attacks and insufficient expected functions.