Opinion | Mr.K Chief Writer | Wendy.L Editor | Emma
Source | Technical Leadership (ID: jishulingdaoli)
What is the hottest thing in Shanghai in the past week?
If the Shanghai Auto Show held at the National Convention and Exhibition Center is said to be the second, I am afraid no one will dare to say the first.
More than 100 years after the birth of the automobile, the 20th Shanghai Auto Show will be held this time. The exhibition area will exceed 360,000 square meters, and the number of visitors will reach one million. The rise of the Chinese market has accelerated the exchange of positions between fuel vehicles and electric vehicles. People have quietly discovered that at this auto show, the status of Chinese and foreign auto companies has begun to change.
Among the more than 1,000 exhibitors at home and abroad this time, Chinese people are no longer superstitious about foreign brand-name car companies. Instead, domestically-made new energy vehicles, as well as related Internet, Internet of Things, artificial intelligence technology and upstream and downstream companies have attracted much attention. He said that the focus of this auto show is no longer limited to "one car".
Nowadays, new energy vehicles with driving automation functions on the market are more common at the L2.5 level, but Baidu and Didi, such as L4 autonomous vehicle companies, have therefore become more and more the focus of attention.
Figure: Autonomous Driving Classification
Since Baidu has been deeply involved in autonomous driving for ten years, it is relatively low-key at this auto show, without too much publicity, and there is not much effort. Strength speaks, after all, everyone knows everything.
What surprised me this time was Didi's autonomous driving, which released a series of new developments in autonomous driving in one go.
For example, this future service concept car Neuron:
Some people say that the silence of driverless rivers and lakes has now been broken by Didi—the reason is very simple. After all, other people have to build a travel platform before landing Robotaxi, but Didi naturally has a travel scene, which is equivalent to being born in Rome. that player.
At present, the number of Didi's self-driving fleet has exceeded 200, scattered in Shanghai, Guangzhou, Beijing, Suzhou and other cities.
When Baidu heard this, it smiled unkindly.
On April 19, on the occasion of the sixth anniversary of the opening of Baidu's Apollo platform, the country's first fully unmanned autonomous driving operation report-"Journey to No Man's Land: 2022 Baidu Autonomous Driving Travel Service Annual Report" was released.
In the report, Baidu announced the latest operating results of its self-driving travel service platform "Carrot Run", with a cumulative order volume of over 2 million, making it the world's largest self-driving travel service provider.
At present, Carrot Run has covered cities such as Beijing, Shanghai, Guangzhou, Shenzhen, Wuhan, Chongqing, etc. In the past year, 7 new commercial cities have been added, and cities with fully unmanned autonomous driving services have covered Beijing, Wuhan, In Chongqing and other cities, the number of self-driving taxis in demonstration operation in Beijing alone exceeds 200.
This fact shows that the arena of autonomous driving has never been silent.
Except for leading companies such as Baidu and Huawei, the research on L4 autonomous driving has been very stable. The other two years have only seemed to be calm on the surface, but in fact they have been turbulent.
There are not only grassroots upstarts who have obtained huge financing, but also high-end self-driving companies with strong eyebrows and big eyes that have suddenly closed down.
Some companies have explored a new way of commercialization, while others have burned money and entered the ICU.
It can be said that several families are happy and some are sad.
01
Want to land L4 autonomous driving? not so easy
On October 26, 2022, an American self-driving company named Argo AI (nicknamed "Two Dog") announced its bankruptcy.
Caused a commotion in the industry.
Originally, it is not uncommon for the autonomous driving track to disappear/appear several grassroots start-ups every once in a while, but Argo AI is different. It is the result of two traditional powerhouses in the automotive industry, Ford and Volkswagen, working together on the autonomous driving track. Said to be the "rich second generation" born with a golden spoon in his mouth.
Combined with the "fracture" listing of Mobileye, the super golden unicorn in the autonomous driving industry, the future Chinese CEO of Tucson was deposed by his own company, and Pony.ai's secret layoffs were followed.
It has inspired the industry's speculation that domestic L4 autonomous driving companies "burn countless money and have no hope of landing".
So all of a sudden, they decided to "reduce the dimensionality" of L4 autonomous driving capabilities and embrace L2.5 instead.
Some people say: Only doing L4 autonomous driving is a dead end. The "impossible triangle" encountered by L4 autonomous driving companies: data, safety and cost are mutually restrictive and cannot be achieved at the same time.
Is this really the case? In fact, it is not the case.
It is not impossible to break through technically, but there are more hidden concerns about legislation, ethics, management and other issues of autonomous driving, as well as the lack of supporting infrastructure for the time being.
In general, the implementation of autonomous driving technology is mainly faced with the following obstacles:
1. Technical issues: Safety is the first priority. Even if there are 0.01% of the scene problems that have not been resolved, it is still not completely reassuring
The realization of autonomous driving technology requires the support of a large number of core technologies such as artificial intelligence algorithms, sensor technology, and control technology. At present, autonomous driving technology is still in the early stages of development, and the maturity and reliability of the technology need to be further improved. One of the most important difficulties is the safety and reliability of autonomous driving. A self-driving car is still a "kid who can drive", even if one day it can handle 99.9% of driving scenarios, but for those low-probability scenarios that have never been learned, I am afraid that it will not respond as well as an "old driver". Once the system fails or malfunctions, it may lead to accidents.
2. Cost issue: autonomous driving, the fragrance is really fragrant, and the price is really expensive.
We mentioned earlier that the implementation of autonomous driving requires absolute safety and reliable performance in terms of technology, but the question is, what is needed for technology research and development? burning money
The cost of research and development of self-driving cars is very high, requiring companies to invest a lot of capital and human resources.
An early self-driving car cost millions.
At the same time, the operating costs of self-driving cars are also very high, requiring companies to continuously optimize algorithms and planning in different scenarios to improve the efficiency and safety of self-driving cars.
Then the problem arises. A taxi driver in a first-tier city earns only 8,000 yuan a month, or about 100,000 a year.
Good guy, the cost of this self-driving car is directly enough to hire a driver for a lifetime.
If the cost of artificial intelligence is higher than that of human beings, how many capitalists will embrace intelligence instead of artificial intelligence?
3. Legislative issues: there is no supporting system of laws and regulations, everything is a mirror image
The promotion of any industry must be supported by sound laws.
At this stage, autonomous driving is an emerging industry, and its laws and regulations have not yet been perfected.
However, there are many legal supports for the demand for autonomous driving.
For example, the question of responsibility determination, who will bear the responsibility after a traffic accident occurs in a self-driving car?
Insurance purchase question, who will purchase insurance for self-driving cars?
There are also a series of ethics, privacy and other issues involved: for example, when you encounter a driving scene, you have to choose which side to go, kill one person, or kill 50 people... Humans cannot make decisions, so what about self-driving cars? Walk?
Just thinking about it makes my brain hurt.
At this stage, self-driving taxis are basically operating on certain road sections with a "test" temporary license. Sooner or later, if self-driving taxis want to have a formal identity on the road, they must solve all these problems. legislation on a series of issues.
02
The arena of autonomous driving:
Baidu continues to reduce costs, Didi crazy speculation concept
What is the ultimate goal of autonomous driving? ——Remove human driving and improve traffic efficiency.
We have analyzed before and found that it is extremely difficult to achieve this goal, that is, the landing of L4 automatic driving.
Enterprises in the self-driving arena have really gone to the extreme to solve these problems.
Security issues first:
A few years ago, Baidu Apollo encountered such a problem in autonomous driving. How to balance safety or riding experience during research and development?
Baidu decisively chose to ensure security.
After all, it's not a big deal if you ride comfortably, but it's a big deal if you get into an accident!
After ten years of hard work, it seems that there has never been a major accident involving Baidu's self-driving taxis.
Second is the cost:
When it comes to money, this is a very real problem.
Where does the commercial appeal of self-driving taxis come from?
It must be cost reduction. You have to make the cost of unmanned driving lower than that of people, so that the market will buy it.
Ever since, cost reduction and efficiency increase have become the most efficient path for commercialization.
And Baidu has gone to the extreme on the road of "cost reduction".
RT5 (also known as Apollo moon) in 2021: The mass production cost price is 480,000 yuan, leading the entire industry towards low cost (involution).
And RT6, which will be released soon in 2022: it will reduce the cost to 250,000 yuan, which is one-tenth of the average level of the entire industry!
There is no most volume, only more volume.
According to the industry, it really can't move
Compared with Baidu's excellent technical strength, Didi chose to find another way and launched the first future service concept car "didi neuron".
According to the introduction of the staff, didi neuron takes "empathy and resonance, everywhere" as the concept, is a "travel tool completely built for passengers", and hopes to provide passengers with ubiquitous services.
In terms of space layout, Neuron canceled the driving position, maximizing the liberation of the seating space.
Under the premise that the total length of the vehicle remains unchanged, compared with ordinary vehicles, the internal space has increased by 50%, and the leg space has increased by 86%. Passengers can even "lay flat" in the car.
In terms of functions, both inside and outside the car have their own highlights: inside the car, the original in-car robotic arm is a "waiter" in the era of unmanned driving, which can provide users with services such as placing luggage, delivering water, and waking up.
The equipped large-screen interactive system is designed with conference, game, audio-visual, street view introduction, atmosphere selection and other modes, which can meet different riding needs.
Outside the car, the traditional car lights are innovatively replaced by intelligent interactive five-link screens, which can more intuitively and conveniently display the status of vehicle turning, arrival, charging, etc.; the design of top wing door + side sliding door improves the safety of opening and closing the door At the same time, it increases the comfort of getting on and off the car.
We just said that Baidu chose safety on the issue of "safety or ride experience". That is to say, compared with those bells and whistles, Baidu's automatic driving will still focus more on a car for travel.
However, Didi's neuron concept car began to focus on the "ride experience".
This reminds me that in October 2020, the German Erwin Heimer Group showed everyone a self-driving concept station wagon called Galileo.
In the future driverless RV, the carriage is equivalent to a mobile hotel, which can directly camp in the wild.
The Galileo concept vehicle is not just a simple camper van, but a complete ecosystem that can provide various services as needed.
We can imagine that one day, autonomous driving will develop into the ultimate form, that is to say, L5 level of completely unmanned driving. Then, the function of autonomous driving cars will not only be as simple as a travel tool, but also include human food, clothing, housing, transportation and work. and so on for all functions.
On that day, when the chat GPT moment of autonomous driving really comes, industries such as transportation, tourism, real estate, and even the entire social organization and lifestyle will undergo subversive changes.
As for when this moment will come in the future, we have to wait and see.
After all, the end of self-driving cars is not only cars, but everything.
(The above pictures are from the Internet, invaded and deleted)
Profile
Introduction of the account owner: Mr. K , Huang Zhekeng, a well-known IT management expert in China, the founder of "Duwu Hill" consulting, and the founder of the technology media "Technical Leadership". The author of the best-selling books "The Way to Cultivate a Technician" and "The Top of Technology Management". He once served as the technical VP of Yiyao.com, the technical director of Yihaodian, and the CTO of Haier's agricultural e-commerce. Share individual growth, team management, enterprise digital transformation, and technology reviews.
-- END --
Send you a copy of "Top Institution ChatGPT Research Report" pdf
Follow the official account below and reply in the dialog box: gpt