Recently, the California Transportation Authority (DMV) released the 2021 autonomous driving road test report, and Chinese and American players continue to lead the industry.
Specifically, the report shows the performance of 28 autonomous driving companies from around the world over the past year (December 1, 2020 to November 30, 2021), including test miles, takeover miles, and fleet size. Among them, the cumulative total road measurement mileage of autonomous driving in 2021 exceeded 4 million miles, doubling from nearly 2 million miles in 2020.
Domestic players perform well
As a bellwether for the autonomous driving industry, the core measure of the report is the annual average takeover mileage (MPI) of autonomous driving players, which is more objective and quantifiable than the test ride experience based on subjective feelings.
Around this indicator, the new intelligent driving combed the overall performance of the 13 most concerned autonomous driving companies in China and the United States, as shown in the following figure:
From the perspective of the number of takeover miles, there are five of the top seven companies in China, namely AutoX, Didi, Wenyuan Zhixing, Yuanrong Qixing, and Xiaoma Zhixing. At present, these players have carried out Robotaxi operation activities for the public, and are gradually moving towards the commercial landing stage.
AutoX, which ranks first among them, has accumulated a large number of autonomous driving scenarios through large-scale, high-density first-tier city operations. At present, it has a R&D technical team of more than 1,000 people, covering the five major R&D centers in Beijing, Shanghai, Shenzhen, Guangzhou and Silicon Valley.
Didi Autonomous Driving, which ranks second in China, has obtained autonomous driving road test qualifications in Beijing, Shanghai, Guangzhou, Suzhou, California and other places, and the team size has further increased to 700 people.
But in terms of overall test miles, Waymo continues to be far ahead, with more than 2.3 million test miles, accounting for more than 50% of the total annual test miles; followed by Cruise, which reached 900,000 miles, and Pony Zhixing from China ranked third, with more than 300,000 test miles.
The test miles of the above three companies account for almost 88% of the total annual test miles.
At the same time, the report also ranks from other dimensions such as fleet size. The top ten companies in California by fleet size, as shown in the following chart:
In terms of data, Waymo's test fleet in California is the largest, with nearly 700 vehicles; cruise and Zoox, ranked second and third, have 138 and 84 vehicles, respectively; followed by AutoX and Xiaoma Zhixing from China, both with about 40 vehicles.
It is understood that the top ten players in the above test fleet are equipped with a lidar, the number ranges from 3-8, and most of them are rotary lidar. However, both Xiaoma Zhixing and YuanRong Qixing have stated that they will gradually switch to solid-state lidar in the future.
As the development of autonomous driving continues to move towards the mass production stage, solid-state lidar can be designed to be highly integrated, lightweight and aesthetically pleasing, and easier to integrate with the original car.
It's important to note that DMVs have mandated that manufacturers testing self-driving cars are required to report any collisions that result in property damage, personal injury, or death within 10 days of the accident.
As of February 1, 2022, DMV has received 405 autonomous vehicle crash reports. Among them, there were more than 100 autonomous driving accidents in the whole year of 2021, while Wenyuan Zhixing from China had 3 cases, Xiaoma Zhixing had 2 cases, and all other Chinese players had no accidents.
In addition, as of November 19, 2021, a total of 7 companies have been licensed to california DMV fully driverless road testing, and players from China include Baidu, AutoX and Wenyuan Zhixing.
Only takeover heroes?
Although dmv's autonomous driving departure report is authoritative in the industry, it has begun to receive more and more controversy in recent years. One of the points of contention is:
The objective complexity of the vehicle operation scenario, and even the subjective judgment of the safety officer, may affect the number of manual takeovers.
This can't help but make people wonder whether it is really objective and reasonable to rank the strength of major autonomous driving players with the cold number of takeover indicators (for details, see the previous report of the new intelligent driving "Automatic Driving, Don't Talk About "Takeover" Color Change").
It is true that the number of takeovers is closely related to technical safety, and the stable handling of common traffic scenarios is the basic literacy of mature automatic driving systems, and only by not relying on safety officers and any form of manual direct intervention can autonomous driving technology bring greater commercial value.
However, it should be noted that the main channel problems encountered by autonomous driving on the road (including known dangerous scenarios) are still far from being fully solved. In different scenarios, even the number of takeovers of the same autonomous driving system can vary greatly.
Some insiders have told the new intelligent driver that if it is a non-pr link (such as the research and development testing stage), players are even more inclined to choose more complex scenarios to hone the system - the meaning of testing is to find problems, and only by constantly finding problems and solving problems can we promote the system to enter a stable iteration period.
Therefore, the frequency of manual takeover may not be simply equated with the immaturity of automatic driving technology, and the takeover rate is not suitable as the only criterion for evaluating the excellence of automatic driving systems.
More importantly, autonomous driving has gradually entered the stage of commercial application. According to McKinsey forecast, driverless cars can generate 200 to 1900 billion US dollars by 2025; Goldman Sachs estimates that the scale of the autonomous driving industry is expected to exceed 600 billion yuan by 2050, and the development process will show exponential growth.
That is to say, in the foreseeable future, autonomous driving will be further integrated into people's lives. As a technology serving human mobility, its system performance, cost control, production line management, and even user experience may need to be deeply optimized.
Leifeng networkLeifeng networkLeifeng network