Are Baidu cars and RoboTaxi good for autonomous driving? No, good for Moutai

2021-02-18 16:02:19

Yunzhong From
Aofei Temple Qubit Report | Public Account QbitAI

The financial report performance exceeded expectations, the stock price rose 3.75% after the market, and the autonomous driving business is becoming more and more promising...

This is a series of new performances of Baidu after the Chinese New Year of the Ox.

In Baidu’s latest fourth-quarter and full-year financial report, Baidu’s revenue and net profit exceeded market expectations, and its non-advertising revenue increased by 52% year-on-year. AI innovation represented by intelligent cloud, intelligent driving and cutting-edge business Business is becoming a new engine driving Baidu's mid- and long-term growth.

In particular , breakthroughs in the field of intelligent driving were not only highlighted by Baidu Chairman Robin Li, but also received additional attention from Mizuho , an internationally renowned investment bank .

Mizuho recently raised Baidu's target price sharply, saying that Baidu's autonomous driving business needs to be revalued-it has been raised to US$40 billion .

And this kind of "reassessment" is naturally not groundless. Nowadays, the three-legged pattern of the world's unmanned vehicles can not be more obvious.

After the annual reports on autonomous driving in China and the United States were released, two official authoritative reports, Transocean Radiance, outlined the latest global autonomous driving landscape.

Waymo of the Google camp, Cruise of GM, and Apollo of Baidu are the top three in the world.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

And it needs to be explained that Baidu Apollo here is only its autonomous driving direction business. Because of Baidu Apollo’s three major businesses, in addition to RoboTaxi as the target autonomous driving business; there are also AVP and ANP that help mass-produced cars achieve smart driving-a more suitable benchmark player on the same platform is Tesla FSD; and smart transportation In terms of direction, it is the only one in the industry.

In addition, in this earnings conference call, Baidu also revealed that the CEO and brand name of the car company have been determined. Li Yanhong said that the reason why he builds his own car is to bring advanced intelligent driving technology to the market as soon as possible and bring the best experience to users.

All in all, let's start with the two reports.

Horizontal Interpretation of the China-US Report

The official U.S. self-driving report comes from California DMV, the first organization in the world to issue self-driving road test licenses and permit self-driving on the road. Naturally, it has attracted the most driverless car players to test and compete.

Its annual statistical report has undoubtedly become the best reference for observing the core progress of each company under a unified dimension.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

What's interesting this year is that China's only official autonomous driving report, the Beijing Autonomous Driving Annual Road Test Report, was released earlier.

So different from previous years, the two reports formed a synergy and resonance this year.

Although the indicators and dimensions are not exactly the same, the two reports in Beijing and California are focused on three aspects .

Road test mileage;

Take over the data;

License issuance progress.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

The road test mileage mainly reflects the amount of the road test. Generally speaking, the longer the actual road test mileage, the richer the scenes that have been seen, and the more long-tail challenges that have been successfully dealt with, the more powerful and safer it is. .

In 2020, California's road test mileage is very different. The first two Cruise and Waymo, relying on their size advantages, will be the best.

Among them, Cruise ran 770,049.26 miles throughout the year, equivalent to 1,239,274 kilometers.

Waymo ran 62,8838.5 miles, which is about 1.0117 million kilometers.

California's third place Pony Zhixing, 225,496 miles, or 362,900 kilometers.

Some people may ask, what about Baidu Apollo?

In this year's California DMV report, there is no drive test data from Baidu Apollo. There may be two reasons.

First, after winning the best results in road testing and taking over mileage in California last year, Baidu said that the focus of road testing will be returned to the country, because RoboTaxi will be launched and trial operations will be launched, allowing higher-dimensional drive testing.

On the other hand, Baidu is oriented towards more advanced technology research and development in the United States, seeking unmanned testing for autonomous driving , and was later awarded a fully unmanned license, but the data in this regard has not yet been released.

So Baidu's drive test data, is it impossible to observe it?

no.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

You can see the Beijing road test report. In 2020, in Beijing, Baidu Apollo's self-driving road test ran 1.1253 million kilometers in the whole year ...

Between Cruise and Waymo.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

Although Baidu is a bit disadvantaged-after all, in addition to Beijing, Baidu has public road test in Changsha, Cangzhou and other places, and California and Beijing have completely different access requirements for drive test licenses.

But whoever makes the official report only have two copies, Baidu Apollo can only count according to the official report data.

So on the road test mileage, the Chinese and American reports are put together. In fact, the pattern is very clear. They are Cruise, Baidu Apollo and Waymo.

Taking over data , whether autonomous driving is enough for old drivers, taking over data is the key indicator currently measured.

California DMV is also the pioneer of annual average takeover mileage (MPI, takeover mileage).

Waymo proactively reported 21 takeovers throughout the year, mainly urban streets and a small number of high-speed scenes, which can be converted to an average of 48,000 kilometers.

Cruise recorded 27 times, also focusing on urban street scenes, which was equivalent to 45,000 kilometers and took over once.

Two Chinese companies followed, AutoX and PonyAI.

However, with the continuous development of autonomous driving, California DMV's MPI indicators are also undergoing a more objective review.

Because MPI data is related to the total mileage of the road test, the takeover style of each company, the road conditions and the difficulty of the scene, it is highly subjective.

Therefore, starting from last year, comparing the data of various MPIs horizontally, they began to encounter the "boycott" of players including Waymo and Cruise. This year, DMV officials also began to restrain themselves.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

The Beijing road test report, from the very beginning, has taken the best of the advantages and disadvantages of California road test. Therefore, in terms of data takeover , the scene and process of the takeover are more important than the result data.

In the road test report in Beijing this year, the focus was on the reasons for the takeover, in which abnormal positioning and perception errors and omissions accounted for the bulk of the technology. On the other dimension, in the key detachment, takeovers during the interaction of traffic participants accounted for a large proportion.

Therefore, in terms of data takeover, the two reports of China and the United States are actually difficult to directly correspond horizontally.

But combined with the license data , maybe it will do.

California and Beijing have two different drive test access modes.

California is liberalized. As long as you apply for it, you can basically get a road test with a safety officer. If you reach a certain level of road test mileage and takeover, you can apply for a safety officer.

Beijing is even more stringent. Above California's conditional threshold, it also has to take an exam, and only after passing the exam can the license be issued.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

The road test difficulty in Beijing is divided into 4 levels (R1-R4). Currently, only Baidu Apollo is conducting R4 level testing.

In addition, only Baidu Apollo has obtained the national open road test qualification certification (Beijing T4 license) with the highest technical level, the highest standard, and the most difficult test scenario in the country.

Of course, the Beijing road test began to have the biggest new change in 2020, and began to issue unmanned licenses. Currently, the first batch of 5 licenses have all been issued to Baidu Apollo, which has passed the unmanned special technology test.

And California has also begun to issue unmanned test licenses.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

A total of 6 companies have obtained it, Waymo and Cruise are both listed, and Baidu is the only autonomous driving company that allows unmanned testing in both China and the United States.

So if the two road test reports from China and the United States are put together, from the perspective of "AI driver level", it may still be the top three pattern on the head.

The top three patterns of global autonomous driving

In a nutshell, it is:

Google camp Waymo, GM's Cruise, Baidu Apollo.

In fact, whether it is the California DMV road test report, China's only official annual report, or the issuance of autonomous driving licenses at different stages, they all show and reflect this pattern.

But maybe there are some friends who don't want to stop at the macro observation, but also want to find the latest "truth" of the global autonomous driving development from these two official reports every year. How should they view these two reports?

In fact, they can complement each other to examine.

The autonomous driving report in the United States, on the one hand, is that there are many participants, and the policy access time and threshold are relatively low, which can be viewed as a “frontier”.

For example, unmanned thresholds and policies. As some companies get licenses, more and more data will be generated in the future, which can be focused on and referenced.

The Beijing road test report is stricter than the California DMV in terms of professionalism, granularity, and entry barriers.

So if you just look at drive test data and subdivided technology progress, China’s only official report is actually more informative.

In addition, at the end of the interpretation of these two annual reports, the key dimensions of the next stage of autopilot racing can actually be found.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

One is manned, and the other is unmanned.

Manned, namely RoboTaxi trial operation and testing, points to the landing of the ultimate product of autonomous driving, which can provide more user experience and reference for future commercial landing.

Unmanned, that is, to go to the safety officer, the inspection is the ability and level of the autonomous driving company to let the "AI drivers" take jobs in all directions.

In these two dimensions, the qualifying competition has also started. Players who have obtained the license actually mean that they have entered this new racing track.

So if you want to participate in the next stage of racing, the key indicator is to look at the license.

So at this stage, combined with the two official annual reports, what is the status of Sino-US licenses?

In California, the licenses for manned tests are issued by CPUC, and currently 7 companies have obtained licenses:

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

Waymo and Cruise are both listed.

In China, Baidu Apollo is the company with the most manned road test licenses, and has obtained manned test licenses in Beijing, Changsha and Cangzhou. Up to now, the Apollo autonomous driving test fleet has exceeded 500 vehicles, obtained 2,900 patents, tested a total of more than 7 million kilometers, and obtained a total of more than 200 test licenses, including more than 150 manned test licenses.

In terms of unmanned operations, California DMV issued licenses to six companies.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

Waymo, Cruise and Baidu Apollo are all listed.

Baidu also took the lead in obtaining an unmanned test license in China.

Therefore, in the next stage of the competition, these three are still the most noteworthy top players and seed players.

After them, Chinese companies such as PonyAI Xiaoma Zhixing, AutoX, Wenyuan Zhixing, and Silicon Valley companies such as Zoox and Nuro have also entered a new stage of the track.

However, if you look at it geographically, players in China may have a first-mover advantage in the new stage .

Because the impact of the epidemic continues, but thanks to China’s victory in the fight against the epidemic and the fastest resumption of work and production, Chinese companies will obviously be on the road faster-it can be proved that Silicon Valley has not been able to resume manned testing throughout 2020. .

Therefore, the anti-epidemic advantage is the time. And China's higher density urban population gathering and scenes are also geographically advantageous for manned and unmanned tests.

With the right time and the right place, in the end there will only be harmony between people.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

At present, the harmony between external and internal people cannot be more obvious.

Baidu, among the top three, has a new high stock price and a market value of 100 billion US dollars, which is the best proof of this kind of harmony.

It is worth noting that Baidu's intelligent driving has repeatedly brought about reassessment by its achievements and strength, and more of it lies in the self-driving business itself.

Because friends who are familiar with Baidu's layout in the field of intelligent driving know that autonomous driving is only one-third of Baidu's part of business.

With the exception of RoboTaxi, which is in the same direction as Waymo and Cruise, the autonomous driving business is the target.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

Baidu also has the AVP and ANP businesses of face-to-face vector production vehicles, which are directly benchmarked with Tesla FSD.

Are Baidu cars and RoboTaxi good for autonomous driving?  No, good for Moutai

 

As well as the "ACE Traffic Engine" on the intelligent transportation track, it integrates Baidu's years of accumulation in AI, big data, autonomous driving, and vehicle-road collaboration. It is the only next-generation intelligent in the industry that is "future-oriented and compatible with the present" Traffic solutions. It is also an unattainable track and direction for Waymo and Cruise-at least in the latest round of Cruise's $30 billion valuation led by Microsoft, this is just an outlook.

So even the latest valuation of US$40 billion is only for Baidu's autonomous driving business. For Baidu Apollo as a whole, it needs to be different.

But in the end, it was all good for Baidu. As Li Yanhong said in the latest full letter:

This era rewards climbers, and this era favors dream makers. This is an era that re-crowned technological innovation. We love this era, let us stand on tiptoes in the deep gully of technology, look a little further, look into the stars, and see the future.

Of course, the acceleration of autonomous driving is not only good for Baidu.

After all, in the long run, the realization of autonomous driving no longer requires human drivers, and there is no challenge of drunk driving, so it will benefit Moutai in the end. (Manual dog head)

What do you say?

- Finish-

Guess you like

Origin blog.csdn.net/weixin_42137700/article/details/113934365