Under the epidemic, to assist in the delivery of medical supplies, the realization of takeout, courier contactless distribution ...... automatic driving technology seems to have more landing scenes. However, just as some of the plucky automotive companies are dialing up the mass-production level of autonomous driving from L2 to L3, smart cars are experiencing a certain sluggishness on the road to commercialization on a large scale at the L3 level.
Earlier this year, Audi revealed its plan to "cancel the L3 autonomous driving research and development program". Although Audi deliberately emphasized? "The team has only shifted to the development of L2 and L4 level autonomous driving technology." But this L3 self-driving system, which was on board the fourth generation of the Audi A8 in July 2017, amounted to finally being obliterated and undeliverable to users after five years and hundreds of millions of dollars.
Previously, the president of Ford (F.US) North America also said that Ford will give up the L3 level of automated driving; Bosch has repeatedly postponed the landing of TJP in public speeches. "L3 landing difficult" is gradually becoming the industry **** knowledge.
The paradox of "human-vehicle takeover timing"
Before answering the question of how difficult it is for L3-level autonomous driving to get off the ground, let's first revisit the history of autonomous driving grading.
In the International Society of Automotive Engineers (SAE) automatic driving classification, the L0-L2 stage belongs to the category of intelligent driver assistance systems, the human driver still needs to be responsible for all the driving tasks, once there is a traffic accident, the responsibility is still borne by the human driver.
In the international automotive engineers (SAE) in automatic driving grading, L0-L2 stage belongs to the category of intelligent driver assistance systems, human driver still needs to take charge of all driving tasks, once there is a traffic accident, the responsibility is still borne by the human driver.
So L3 is caught in the middle of it all, and it looks very awkward. Specifically, L3-level autonomous driving requires the human driver and the intelligent driving system to frequently hand over control of the vehicle during use, and the technical realization of this human-computer interaction process is very complex and contingent.
Because it is in the gap between human to vehicle excess, L3 level autonomous driving does not have a clear division of responsibility between the vehicle and the human driver, and in the event of an accident, the subject of responsibility may involve multiple parties such as the driver, the OEM, and the software provider.
When we look back at the Chinese market, however, we find that new L3-level autonomous driving vehicles have already hit the ground running. Changan Automobile's UNI-T and SAIC-Rongwei's MARVEL-R are even fighting for the title of the world's first 5G mass-production model with an L3-level autonomous driving system. In addition, a number of domestic autonomous car companies are planning to realize the mass production of their L3 models in 2020. These include Great Wall, Geely, Chery, Dongfeng, GAC New Energy, and Red Flag.
Why is there such a big difference between domestic and foreign attitudes toward L3? The main reason has to be explained from the perspective of "legislation". In addition to the global SAE classification standard, China ushered in its own automatic driving classification standard in March this year. The Chinese standard is roughly the same as the world standard, but there are still some differences.
The biggest difference is that the definition of L3 in the Chinese version of the classification standard explicitly adds requirements for driver takeover monitoring and risk mitigation strategies. However, when it comes to enterprises, our L3 standard has to a certain extent alleviated the "burden" and "safety concerns" of automobile enterprises. That's why domestic automakers are taking a different trajectory than multinationals.
If you think about it from the consumer's point of view, I'm sure some of the foreign car owners will be willing to give it a try? The typical L3 usage scenarios of "hands off", "feet off" and "eyes off". Tesla owners have mistakenly thought that L3 Autopilot could be completely free of human intervention, and have ended up in car accidents. The same thing almost never happens to Chinese consumers. For us, "It's my business if I use it or not, but whether you deserve it or not is another matter." But if you want Chinese car owners to really put their lives in the hands of AI, I believe no one will do so. This kind of Chinese and foreign consumer psychology of the essence of the difference, but also caused the current many independent brand models of the "intelligent configuration" is far higher than the joint venture and luxury brand models of the "strange phenomenon".
The author believes that this is a watershed opportunity for independent brand enterprises to realize full leadership in the technical level.
Without L3, trying to jump directly from L2 to L4 is just an impossible thing.
The L3 level can serve as a transition, both to show the industry and consumers the technology, but also to prove that they are able to develop high-level autonomous driving systems. Jumping directly to L3 autonomous driving means that car companies will lose this part of the market share, resulting in a strategic and comprehensive backwardness in the level of technology.
2020, the first year of L3 landing?
Watching everyone's L3 models have landed, many people call 2020 the first year of L3 autonomous driving, in fact, this is not the case. Enterprises can make production models equipped with L3 in advance, but whether they can use all the functions of L3 autonomous driving without restrictions in the actual car scene depends on whether laws and regulations allow it or not.
Zhu Huarong, president of Changan Automobile, also specifically mentioned the policy restrictions on the day of the release of the L3 level of autonomous driving, saying, "After the road conditions and policy factors allow, the fastest Changan Automobile L3 level of automated driving technology can be realized within the year of mass production."
It can be seen that for the development and large-scale application of automatic driving technology landing, the country has not introduced any key to allow the use of L3-level automatic driving vehicles pre-conditions.
But the Chinese version of the Automated Vehicle Driving Classification is a good start. At present, from the national policy as well as the industry's **** knowledge, L3 level autonomous driving or even L4\L5 autonomous driving to really early arrival, vehicle-circuit coordination is the only path to realize autonomous driving. This means that not only the single car has the intelligence, more importantly, the road and related infrastructure can also be "smart" up.
The three core components of the VRD system are: intelligent vehicle system (vehicle side), intelligent roadside system (roadside + cloud) and communication platform. This means that the intelligence of the vehicle, the intelligence of the road and the network connection between the two, forming a three-dimensional architecture, can greatly enhance the ability of automatic driving, and even the deployment of low-cost induction equipment in the vehicle end, you can make the vehicle with a certain degree of automatic driving ability, greatly reducing the technical and cost threshold of the self-driving car.
Vehicle-circuit coordination also requires the fusion of single-vehicle sensing data from different vehicles, and the characteristics of data from different sources vary greatly. This requires the in-vehicle OS to perform multi-level information fusion at the data level, feature level, and decision level to achieve a higher level of integrated decision making. This requires a unified OS system with high reliability, high compatibility, and high-level information fusion, and the realization and application of this system needs to be supported from the cloud.
In this sense, 2020 will not be the first year of L3 autonomous driving, and it may not even be realized in the next two years. But we'll still have partial access to the upgraded driving experience that L3 technology brings, and that's a good thing for car owners.
As for the commercialization of the scale of the application, we can just think about it for the time being, and don't take it too seriously.
This article was written by the author of Automotive Home, and does not represent the viewpoint of Automotive Home.