Source | BohuFN
Author | Chen Ping An
In recent days, the hottest topic in the circle of cars, apart from the Shanghai Auto Show, is nothing more than a big discussion about automatic driving.
Last month, Wang Chuanfu, BYD's chairman, enthusiastically opened the dialogue at the investor communication meeting after the 2023 performance conference, saying that automatic driving was all nonsense, just a new suit of the emperor. Automatic driving was threatened by capital, which has been fooled for years.
"It took me tens of thousands of engineers to figure out whether the road of automatic driving is impassable. BYD has been making right decisions all the way to the present. Can I be wrong this time?"
As soon as the words were said, a stone splashed a thousand layers of waves. It is recognized in the industry that the first half of new energy vehicles is electrification, and the second half is intelligence. Many car enterprises regard automatic driving as the key to the next step of new energy vehicle industry competition.
Soon a big man came out to refute. At the recent press conference, in addition to boasting about his own products as usual, Yu Chengdong, CEO of Huawei's consumer business, also responded to the statement that "driverless driving is nonsense", saying: "There are two reasons for saying that automatic driving is nonsense: either you don't know the industry; or you mean to say that you didn't do a good job, so as to attack the industry."
On the eve of the opening of the Shanghai International Automobile Exhibition, at the 2023 technical architecture conference, He Xiaopeng, the founder of Xiaopeng Auto, said directly in the behind the PPT that "XNGP is the ultimate form of intelligent assisted driving, so that driverless driving is no nonsense."
(Source: Vehicle)
In fact, the controversy about automatic driving has a long history. Last year, the automatic driving industry experienced a cold winter, and it was difficult to finance and land. Argo AI, a star company backed by Ford and Volkswagen, announced bankruptcy directly, which surprised the onlookers.
From the perspective of market reaction, Xiaopeng and Tesla, the engine manufacturers that pay more attention to the research and development of automatic driving technology, did not gain additional sales mark.
Is automatic driving really a big trick?
Is autonomous driving a pseudo concept?
At the previous investor communication meeting, Wang Chuanfu talked a lot about automatic driving, which generally includes the following two points:
1. It is still difficult for the production line to engage in intelligent and unmanned production. Unmanned driving is more complex and difficult to achieve. ADAS algorithm and high-order auxiliary driving are deified under the constraint of capital, and will return to rationality in the future.
2. At present, the assignment of rights and responsibilities for automatic driving is not clear. Once an accident occurs, it is unfair to consumers and brands.
Admittedly, Wang Chuanfu's view on automatic driving is quite accurate.
Based on the degree of unmanned driving, automatic driving is divided into 6 grades from L0 to L5:
L1 and L2 are auxiliary driving and need driver control or take over at any time;
L3 and above are automatic driving, and the main responsibility for the accident has changed from the driver to the system, but L3 still requires the driver to take over in the driving position in a timely manner;
L4 and L5 are truly driverless, and people's time in the car is completely liberated, so they no longer need to pay attention to driving.
For the research and development of automatic driving technology, the industry has developed different paths.
Automatic driving companies on the market usually focus on L4 level automatic driving solutions, but they are generally difficult to solve the contradiction between data, safety and cost.
Companies that do L4 level automatic driving usually face a long tail problem. The current algorithm can only complete the previously trained scene tasks, but it is difficult for your training scene to cover all situations. The scenes that are difficult to predict, even if there is only a 1% probability, are exactly the ones that need to be solved most.
The best way to solve the security problem is to have enough test data. However, the premise for the popularity of L4 automatic driving is that it is at least safer than human driving, so automatic driving companies can only test in specific sections, and the amount of data obtained is relatively limited. By the end of 2022, Waymo, the world's largest autonomous driving mileage last year, had only 2.32 million miles.
Not to mention that in order to ensure safety, L4 level automatic driving is often "over matched" in perception, and the high cost severely limits the commercialization ability of automatic driving companies.
Therefore, high-level automatic driving is still a rather elusive goal. Last year, many "noble" autonomous driving companies had to put down their positions and apply L4 technology to L2+vehicles, first laying down costs, ensuring safety, and then slowly mapping the data.
On the other hand, the OEMs led by Tesla and Xiaopeng have chosen a gradual development model. Although this model of Tesla is regarded as a more promising path by the industry, it has been since the launch of the FSD beta in the second half of 2020, and the official version has not been announced yet.
In fact, the popular automatic driving in the market actually refers to L2 level automatic driving, that is, the high-level auxiliary driving mentioned by Wang Chuanfu, not L3 level, which is known as the watershed of automatic driving to achieve automatic driving in specific environments.
From the policy point of view, although Shenzhen issued the Regulations on the Administration of Intelligent Connected Vehicles in Shenzhen Special Economic Zone in June last year, the laws and regulations against driverless vehicles in case of traffic accidents are not perfect, and the rights and responsibilities are not clear. Car companies are prone to appear when they launch vehicles rashly, as described by Wang Chuanfu, "a car accident will make the brand and the car impossible to sell" Or the phenomenon that only consumers can bear the responsibility.
In addition, some accidents related to automatic driving have also caused users to worry about safety.
From a more utilitarian point of view, automatic driving is not the decisive factor of sales, but price, endurance and safety are the first. In 2017, Wang Chuanfu said in an interview: "During the new energy conversion, it is more important to push some cars that are acceptable to the public."
Tesla FSD has only 7% option worldwide. In China, although there is no official data, only 1% - 2% is estimated by institutions.
However, BYD, which has always been criticized for its backward intelligence and the biggest feature of its own auxiliary driving system DiPilot, is that it sold 1.8635 million vehicles last year. In the first quarter of this year, it beat Volkswagen to the top of the Chinese market in a record.
To be fair, in the case that the relevant laws and regulations are not yet perfect and the technology is not yet mature, lowering the expectation of automatic driving is a more appropriate and stable choice for both vehicle enterprises and users.
Why are car companies still betting on automatic driving?
The above discussion does not mean that automatic driving is not important, and car companies can abandon it as our shoes. On the contrary, since the beginning of this year, many automobile enterprises that once did not focus on automatic driving have focused on automatic driving.
On the first day after the year, Li Xiang, CEO of Ideal Auto, said in a letter from all members that he would become a leading AI enterprise in the world by 2030.
On April 18, Ideal released the "dual energy strategy" at the Shanghai International Automobile Exhibition, which not only announced that ideal intelligent driving has entered the 3.0 era in a high profile, from high-speed scenes to urban scenes, but also announced the landing time:
The NOA of the ideal AD Max 3.0 city will start to push the internal test users this quarter, and will push 100 domestic cities by the end of the year, using a set of technologies to fully connect cities and highways.
He Xiaopeng, the founder of Xiaopeng Auto, mentioned in the letter that intelligence should be established as Xiaopeng's biggest differentiation advantage in the industry.
At the 2023 technical architecture press conference, Xiao Peng mentioned that he has opened the first stage of the full scene assisted driving XNGP to some car owners, and can realize NOA in three open cities of Shanghai, Shenzhen and Guangzhou based on the high-precision map
"By the second half of this year, Xiao Peng will have the ability to automatically change lanes, overtake and turn left and right in cities without high precision maps."
In terms of publicity, Interworld brand appears more radical. At the press conference, Yu Chengdong highlighted the HUAWEI ADS 2.0 advanced intelligent driving system and said:
"HUAWEI ADS 2.0 is infinitely close to L3 in technology, we say L2.99999......"
"Huawei ADS 2.0, the first of Huawei's advanced intelligent driving version in the M5 of AITO, has been infinitely close to L3 advanced intelligent driving. This system can reduce traffic accidents caused by inattention and complex road conditions by 90%."
According to Yu Chengdong, Huawei's advanced intelligent driving system can cover 90% of urban scenes, and has been launched in Shenzhen, Shanghai and Guangzhou. Chongqing and Hangzhou will also be unlocked in the second quarter of this year. By the third quarter of this year, Huawei's NCA in urban areas will have 15 cities without maps, and another 30 cities without maps will have landed in the fourth quarter, reaching 45 cities.
Even BYD has made many moves in automatic driving over the years. For example, cooperate with intelligent driving companies such as Monenta and Horizon; BYD will carry NVIDIA's DRIVE Orin computing platform and laser radar in some models of the new generation Dynasty and Ocean series.
In the view of Bohu Finance, automatic driving is not only the key to intelligent competition, but also the key to distinguish new energy vehicle enterprises from traditional vehicle enterprises. In Musk's vision, selling cars is not the main source of profit. In official terms, cost pricing has always been Tesla's pricing principle. The greater imagination comes from the profitability of software, which is the key to the transformation of new energy vehicle enterprises from manufacturing enterprises to technology enterprises.
Write at the end
Since the birth of the automobile, people have never stopped fantasizing about automatic driving.
In 1960, Germany made some attempts related to automatic driving. In the 1980s, the "Navlab" project of Carnegie Mellon University (CMU) gave birth to the world's first vision based autonomous vehicle. Fifteen years later, Navlab 5 ran 2850 miles, 98% of which were self driving.
(Source: Network)
People are eager to liberate themselves and reduce accidents through automatic driving technology. As long as this demand exists, automatic driving is destined to be the most critical competition in the industry.
Reference source:
1. Cyber Auto: High precision map has changed from "Little Sweetie to Mrs. Niu", and car enterprises and graphic merchants have dumped each other
2. Yuanchuan Auto Comments: Unmanned driving, no one alive?
(Statement: This article only represents the author's view, not Sina.com's position.)
