In 2021, the popularity of driverless cars has faded, and "5G cars" are in the ascendant

It’s been a while since 2020, but the promised 1 million self-driving taxis are still nowhere to be seen. The car companies that vowed to make some achievements in the field of driverless cars this year have also stopped one by one.

I don’t know if these experts who once shouted for “unmanned driving by 2020” will have new goals, but reality has taught them to be low-key. However, there are exceptions, such as Baidu’s Robin Li. At the Baidu World Conference 202, he “boldly” predicted: “Unmanned driving technology will definitely enter the large-scale commercial stage in 2025.”

However, the driverless technology he mentioned needs to be put in quotation marks, because it is not really "driverless", but the driver has been moved to another space. According to his original words: "'5G cloud driving' is Baidu's latest progress in driverless technology in the past year. In the past, when driverless cars were tested, there was usually a safety officer in the car. '5G cloud driving' means that there can be no safety officer in the car, because we have configured a safety officer in the cloud to help the vehicle get out of trouble."

5G cloud driving is actually nothing new. Last year, GAC Research Institute’s Technology Day demonstrated related technologies.

Baidu's "5G Cloud Driver" has probably revealed the true "pill" of the driverless system. At least in the foreseeable future, the driver is still indispensable, but he can use the characteristics of 5G instant communication to transfer the driver from the car to the cloud.

Autonomous driving has gradually fallen out of favor, but a new concept is gaining popularity: the so-called "5G car." Today, major automakers are obsessed with 5G cars, and are even vying for the title of "the world's first 5G car."

For example, GAC Aion defined the AION V (parameters | inquiry), which was officially launched in June last year, as the world's first 5G car; the BAIC ARCFOX α-T, which was pre-sold later, claimed to be the world's first commercial model equipped with 5G technology; BYD was not to be outdone, declaring Han as "the world's first model equipped with Huawei's 5G technology"; in addition, Roewe and Great Wall are all emphasizing their "firsts".

Judging from the current situation, Huawei will dominate the 5G automotive field in the future.

I can understand why they are vying for the "first place". After all, we have been instilled with the competitive concept of "only remembering the first place, not the second place" since childhood. 5G technology has a subversive effect on automobiles. It can even be said that it is the key for my country to achieve its transformation from an automobile power to an automobile power. Therefore, competing for this "first place" is of great significance.

"Isn't 5G just about making our internet faster? Why is it so important to the automotive industry?" Many people may be skeptical. Indeed, the 5G phones we use now have no advantages other than faster internet speeds than 4G phones. But in fact, in addition to fast internet speeds, 5G also has two features: low latency and wide connectivity.

High speed, low latency, and wide connection, these three characteristics of 5G give cars two major features, one is instant communication, and the other is the Internet of Everything. These two major features are precisely the two foundations for my country to achieve vehicle-road collaboration in the future. The importance of vehicle-road collaboration can be clearly stated in the words of Wang Jin, CEO of China Intelligent Driving: "Vehicle-road collaboration is an opportunity for China's driverless car to overtake others."

At present, two routes have been developed for autonomous driving. One is the single-vehicle intelligence that the United States prefers, and the other is the intelligent road that my country is vigorously developing, that is, the vehicle-road collaboration mentioned above. I can use an analogy to explain the difference between the two: when the world is full of bad roads, do you buy yourself a pair of better shoes, or repair all the roads in the world?

From an individual perspective, it is of course to buy a pair of good shoes, but from a national perspective, it is of course to repair the roads. The former is the position of American companies - they can only take care of themselves and it is difficult to push the government to do anything; while the latter is the position of our country. After all, we have institutional advantages and the efficiency of promoting infrastructure construction is still quite high.

In fact, in terms of functionality, vehicle-road collaboration is indeed better than single-vehicle intelligence.

Wang Jin mentioned that the intelligence of a single vehicle relies on the computer on the vehicle to make decisions, and tends to optimize individuals. If only individual optimization is considered, the so-called "Bartner effect" in mathematics will be formed, which will inevitably cause congestion. It is impossible to solve this problem by relying solely on deep learning. If the vehicle-road collaboration solution is adopted, there will be a central computing center to coordinate and optimize from a global perspective. The optimization efficiency is much higher than that of a single vehicle, and it can solve the congestion problem from the root, greatly improving the traffic efficiency of vehicle-road collaboration.

At the launch ceremony of the GAC Aion AION V 5G version, the author also experienced the basic version of vehicle-road collaboration, that is, only the vehicles are connected to each other. But even if it was just the basic version, I felt the power of vehicle-road collaboration - the 5G version of the vehicle is like having X-ray eyes, and can sense vehicles that are also connected to 5G in the blind spots of vision, so as to achieve early perception and preparation.

Wang Gang, chief scientist of Alibaba AI Lab, conducted an experiment. At a speed of 30km/h, when a general vehicle encounters a "ghost jump (a pedestrian suddenly rushes out)", there will be 100% casualties. However, with the help of vehicle-road collaboration, they can achieve a 100% successful avoidance rate.

It can be said that once the vehicle-road collaboration is widely implemented, all the current road traffic problems will not be a big deal. The biggest problem on the road at present is disorder. Every road participant is striving to maximize their own interests, while ignoring the overall interests, which eventually leads to global chaos.

In the context of vehicle-road collaboration, everyone on the road is just a point on the network, and is controlled by the "hand of God", so traffic efficiency will undoubtedly be greatly improved. Most importantly, everyone on the road can sense the presence of each other, and the system can predict road conditions in advance, so unmanned driving is a natural thing.

Precisely because of such a big move, GAC Aion boldly declared at last year's Guangzhou Auto Show that it would achieve L4-level autonomous driving functions by 2025; Great Wall mentioned in its latest Coffee Intelligent Driving "331 Strategy" that it would achieve L4-level autonomous driving with the most coverage of Chinese scenarios by 2023.

Achieving L4 autonomous driving in the field of passenger cars is already a remarkable level. According to the SAE classification of autonomous driving, L4 is already a highly autonomous driving level. At this level, the driver is not necessary in specific scenarios, and the autonomous driving system can automatically respond to all situations in these scenarios. In other words, it can already achieve unmanned driving functions in specific scenarios.

I believe that if all road participants, whether people, cars or objects, are part of vehicle-road collaboration, it will not be difficult to achieve L4 autonomous driving function. But the problem lies in the fact that it is quite difficult to popularize vehicle-road collaboration.

Hua Guodong, director of Jiangsu Intelligent Future Automobile Research Institute, proposed six major difficulties in building a vehicle-road collaboration system at the vehicle-road collaboration sub-forum of the "2020 Yangtze River Delta Intelligent Transportation and Intelligent Connected Innovation Development Forum". I will not elaborate on them one by one here. Among these six points, I think the most difficult to solve is the operation model problem: how to increase user penetration?

The government, known as the "infrastructure maniac", does not find it difficult to lay basic equipment for vehicle-road collaboration on the country's more than 4.7 million kilometers of roads. It is just a matter of determination and time. Before that, I hope that street lights can be popularized on highways first, but it will be quite difficult to popularize 5G equipment on nearly 400 million motor vehicles.

Let alone 5G equipment, the convenient and easy-to-use ETC is also difficult to popularize. After the nearly "Great Leap Forward" promotion in 2019, the cumulative number of ETC users reached 204 million (as of the end of 2019), and as of June 2020, the number of cars in use was 270 million. In other words, the penetration rate of ETC is only 76%, which is still a long way from the original target of 90%.

ETC is such a convenient and free thing, but it is so difficult to mobilize them. What's more, you want them to install something that has nothing to do with them and may even harm their interests? You know, there are nearly 400 million vehicles, and they do not have the conditions for autonomous driving. Installing 5G equipment on them is a burden. It is also unnecessary. Installing an unimportant device on them will not only take up space, but also increase the failure rate.

▲According to GAC Aion, the 5G sensing on AION V is currently limited to AION V and AION V. It seems that if you want to experience 5G vehicle-road collaboration now, you can only make an appointment with a friend in the car club and go to the designated test section.