Unable to keep the high valuation of autonomous driving

 The commercialization of self-driving cars is really far from us.

On September 22, a car owner drove the Ideal ONE on Qingdao high-speed. Because of his trust in assisted driving, he rear-ended a van. At the same time, the A-pillar broke and all 7 airbags failed to pop up. A family of three almost died due to automatic driving. .

Afterwards, the car owner issued a soul torture to the ideal car : "Why can't the driving assistance system recognize the lane change of the car in front?" The ideal car was speechless.

Until the second day after the owner issued the document, Ideal Auto responded to the "Ideal ONE Qingdao accident", saying, "Because of the limitations of autonomous driving technology."

However, the answer carefully prepared by Ideal Car has not been recognized by the public. This matter is still being fermented, and it continues to plague Ideal Car.

As we all know, this is not the first time an accident has occurred in autonomous driving. Since the opening of autonomous driving road tests around the world in 2013, there have been several accidents almost every year.

This not only affects the people's trust in autonomous driving technology, but also makes many people feel that autonomous driving is still far from life.

Therefore, the ToB industry headlines also want to ask, where is the high-valued autonomous driving?

01  Two-year trial operation, autonomous driving is on the same track

On the afternoon of October 13, a group of people gathered on the side of an empty road near the Daoxianghu subway station in Haidian District, Beijing, out of step with the quiet surroundings.

Most of them are young people in their twenties and thirties, mostly men. They look like a group of companions going to a party, but in fact, these people are all reporters or IT technicians who come to experience Baidu's self-driving taxi.

A few days ago, on October 10, Baidu officially announced that self-driving taxis will be fully open in Beijing. Beijing citizens can ride at the self-driving taxi stations in Haidian and Yizhuang. You don’t need to make an appointment and you can just place an order for free trial rides. One stone caused a thousand waves of waves, and once the news was released, it instantly “fired” all over the capital.

"Today is the second day of opening. I thought there would be fewer people, but now it seems that my prediction is not accurate." A colleague in the media was talking to me while constantly looking down at the phone and staring at the order on the screen.

However, even with constant attention to order news, Baidu's self-driving taxis are still missing. And every 5 minutes, the original order will be automatically cancelled and you need to call again.

Finally, after 25 minutes, the system successfully took the order. The security officer of Baidu's self-driving taxi called and confirmed the personal information, boarding location and itinerary with ToB industry headlines, and drove slowly in 5 minutes.

This is a black Lincoln MKZ with high-tech equipment such as vision radar, millimeter wave radar, lidar, sensors, GPS, etc. installed around the body. Although it looks bulky, it feels valuable.

After getting in the car, the system prompts that you need to measure your body temperature. At the same time, you must confirm the information by scanning the QR code on the screen directly in front of the seat, and then you can start the journey after you click to start the journey. During the driving process, the taxi will always maintain an automatic driving mode unless it is taken over manually.

After completing some preparations, the self-driving taxi finally started. However, the driving speed of the car is below 40 kilometers per hour, and once it changes lanes, it needs to be taken over manually. It is not as smart as imagined, and it is accompanied by some noise.

"The road speed limit is 40 km/h. Vehicles will never exceed the speed limit. The maximum speed limit of the system is 60 km/h. This is a policy decision. As for noise, it is the sound made by the ventilation system in the trunk, which is required during computer operation. Dissipate heat.” The safety officer in the car explained, and it seems that he has responded to these questions many times.

About 10 minutes later, after the self-driving taxi arrived at the destination and got off the bus, the ToB industry headline once again met the same colleague above. He said: "After this experience, it feels more difficult to commercialize self-driving, because this time Compared with the self-driving taxi he experienced in Guangzhou two years ago, there is basically no difference."

It is understood that the first self-driving taxi in the country launched by the self-driving company Wenyuan Zhixing in Guangzhou in 2018 can not automatically change lanes, and has almost the same characteristics as Baidu's self-driving taxi. In addition, the same is true for the autonomous driving introduced by the automaker Tesla.

ToB industry headlines have conducted experiments. In addition to manual adjustment of the maximum speed limit and no strong noise in the car, Tesla's automatic driving is also difficult to achieve automatic lane change.

According to the ToB industry headline survey, nearly 80% of the autopilots currently available in the country expressed dissatisfaction and did not meet their expectations. It is not difficult to see that in the past two years, there has been no major breakthrough in technology for autonomous vehicles.

A media colleague who has experienced Baidu's autonomous driving even said: "There are no surprises, plain, and no expectation of the experience, the overall is pass, with the expectation of the experience, only disappointment."

02Prepare  the next step to solve the problem, in the road intelligence

"When the real security officer removed, far removed programmable system, is the real autopilot." Autopilot field of investor Lin Meng forthrightly said that the current bicycle manufacturers introduced just a semi-intelligent.

It is undeniable that Lin Meng said the truth. Because from the current technology point of view, just making the car smart cannot achieve the goal of the car being responsible for the driving process.

For example, blind spots and obstructions. Because the viewing angle and height of the vehicle sensor are limited, the vehicle always has a blind spot for driving, and if there is an obstacle on the side of the road, it is difficult to see the situation behind the obstacle.

There is also the perception distance. At present, the effective perception distance of most lidars is about tens of meters. If the vehicle speed is a little higher, there is no time to react when there is a stationary object in front.

In addition, high-precision maps cannot be updated in real time, and complete information cannot be obtained for decision-making and planning. It is a difficult problem that cannot be solved only by the car becoming intelligent, and it is also known as the defect of bicycle intelligence in academics.

To solve this problem, autopilot manufacturers after the test several times the amount of the down, adjust the direction of the research proposed to ensure "smart car", but also let the "smart way."

In terms of specific scenarios, on a driving highway, if a group of driving vehicles can interact with the road all the way, when a car is about to change lanes to overtake, the road will receive relevant signals and give feedback to other vehicles. Road change scenarios, automatic lane changes and other behaviors will no longer have the possibility of traffic accidents.

Just like at a congested intersection, there is a traffic policeman adjusting the vehicle all the time. Intelligent roads act as traffic police. The technical term is vehicle-road collaboration.

In Ali’s words, a higher level of autonomous driving requires not only smart cars but also smart roads.

Ran Bin, Dean of the Southeast University-University of Wisconsin Intelligent Networked Transportation Research Institute, said that if the road and the car are coordinated together, there is no need for such strong artificial intelligence to realize automatic driving, and the third level of automatic driving can be realized quickly , A single car can save 50%-90% of the cost, saving about 10 years of time.

At present, vehicle-road collaboration has become an important direction for many autonomous driving companies, and Baidu, Huawei, Tencent, etc. are all making efforts to this end.

Although such an idea is good, it is also difficult to implement. To make the roads smart, it is necessary to transform the roads, which cost a lot of trillions of yuan.

In addition, "Road Intelligence" requires vigorous promotion from the government to make data available across the country, which also has extremely high requirements for the collaborative work of various local governments.

Even if the above situation can be completed, the time to rebuild the road is also an issue. Just like the establishment of 5G base stations, full construction will take a long time.

Based on the above viewpoints, it is not difficult to see that autonomous driving has a long way to go if it wants to integrate into life scenes.

03  Capital enthusiasm is no longer, valuation turned into a bubble

The slow commercialization of autonomous driving has already disappointed the public. But what is more disappointing than this is that capital is becoming less optimistic about this field.

The data shows that since the 2016 U.S. investment in autonomous driving drove the global investment peaks in 2017 and 2018, the capital market has delayed the original plan in 2019. By 2020, most of the autonomous driving start-ups are already weak.

(Data source: IT Orange)

Data from it Orange shows that about 75% of the current autonomous driving companies have not yet passed the B/B+ round of financing. This indicates that most of the autonomous driving companies that have not yet reached the C round of financing will not receive funding support and will eventually die.

For example, in 2019, Drive.ai, a star self-driving company that was personally involved in the operation of the former Baidu chief scientist Wu Enda, announced its permanent closure. The important reason for the closure was not entirely due to technology. The market summarized it as poor financing.

In addition, in June of this year, Zoox, a foreign leading autopilot company, also sold Amazon due to a "funding chain break", and its purchase price was less than half of the previous round of financing valuation.

Focusing on the head autopilot companies, such as Baidu, Google and some auto OEMs, although they are favored by the capital market and are not bad for money, the technology has not been able to wait for a breakthrough, and eventually capital will lose patience.

For example, in 2014, Apple “smashed” US$3 billion to acquire Beats and secretly formed a self-driving car team, code-named “Project Titan”. However, after five years of arduous research and development, Apple suddenly let go and fired more than 200 "Project Titan" related employees to slow down its investment in self-driving cars.

Similarly, even after 9 years, the field test mileage has reached 16 million kilometers. Waymo, which once again won the safety drivers, is not daring to give the specific time for the commercialization of autonomous driving at will.

Various phenomena have shown that autonomous driving has gradually deviated from the support of capital and moved from high-speed development into a slow lane. The corresponding concepts will also die down and become cool.

This phenomenon is surprising, because just two years ago, Google's Waymo received $175 billion in 2018, which is double the valuation of more than $70 billion last year.

After the driverless startup Tucson completed a $95 million Series D financing in February 2019, the accumulated financing amounted to 178 million U.S. dollars, and the valuation also exceeded 1 billion U.S. dollars.

What is regrettable is that the concept of autonomous driving suddenly collapsed as soon as it became an outlet. These companies are likely to become the once popular "stars", and their experience is really embarrassing.

Therefore, the current hopeful capital for autonomous driving is more gambling. The bet is that self-driving technology will be able to make breakthroughs in the near future, so as to quickly commercialize and realize huge profits.

However, although this possibility exists, the probability is relatively slim. It is even more likely that this breakthrough will be delayed in a few decades, but that is something these capitals cannot afford to gamble on.