Rewrite the Technological Landscape

After Musk left, Meng Qian directly made appointments with BYD's Wang Chuanfu and Geely's Li Shufu. Because time waits for many things, these two companies have been quite active recently.

"I heard that both BYD and Geely have begun to increase investment in electric vehicles."

"This year, Omi vigorously promotes electric vehicles. After the Copenhagen conference, electric vehicles have become a hot topic in the industry. Meng always knows that I have always believed in the future of electric vehicles." Probably because of Li Shufu's presence, Wang Chuanfu's words are obvious keep it.

In fact, the two of them were a little surprised when they found out that they were being asked out at the same time. At this moment, the two companies are gradually becoming equal. They are rivals anyway, even though Dafeng Group is a major shareholder, but Meng Qian said that today they are here to talk about future plans. Competitors Sitting together and talking about future plans, nothing feels right.

Sure enough, Li Shufu just said in a very official way, "We still have a lot to learn from BYD in the development of electric vehicles."

Knowing the little thoughts of the two, Meng Qian went straight to the topic, "Today, when I asked you two to come over, first of all, I want to tell you that I am going to reconcile with Musk."

Meng Qian roughly talked about the matter.

"Tesla's research and development achievements in motor systems this year are indeed very valuable." Regarding Tesla's technology, Li Shufu is quite convinced. After all, when Tesla devoted itself to pure electric vehicles, Geely did not develop much in this direction. .

But Wang Chuanfu was obviously not reconciled, "Actually, we have a lot of patent overlaps with Tesla, and we are also speeding up patent applications recently."

"Actually, it is precisely because of BYD's development that I want to get Tesla's patents over, so that when the two sides are combined, we will have the opportunity to attack more advanced technologies."

The two nodded, agreed with Meng Qian's statement, and then waited for Meng Qian to continue. After all, they didn't know what Meng Qian wanted to talk about when he came to him, so Meng Qian changed the subject directly, "This matter is related to Let me say hello first, the specific situation will be discussed after I reach a final agreement with Musk, and the specific development of electric vehicles will definitely depend on the two of you.

The main reason for me to invite you two today is another thing. This year we saw that electric vehicles will become a trend in the development of automobiles, and we should also see another trend in the development of automobiles. "

Wang Chuanfu and Li Shufu looked at each other, but they didn't think of the answer, and the last topic ended like this, and they were quite confused. Facing the puzzled eyes of the two, Meng Qian gave the answer, "Unmanned driving."

Upon hearing the answer, the two suddenly realized.

"Google has officially deployed driverless cars this year, and it is backed by the U.S. Defense Advanced Research Projects Agency. A large number of R\u0026D personnel from the agency have gone to Google. In this field, I don't want to lag behind." Meng Qian expressed his opinion bluntly.

"What specific plans does Boss Meng have?"

Meng Qian indicated to the two of them to watch the PPT presentation, "Our company has come up with a conclusion that sensors + high-precision maps + cloud computing + rational decision-making algorithms are the most likely solution to realize driverless driving.

Therefore, if you want to develop unmanned driving, you need a corresponding lens, a corresponding CMOS image sensor, a corresponding DSP digital processing chip, a corresponding millimeter-wave radar, a corresponding laser radar and other hardware equipment, as well as an algorithm with rational decision-making ability and a security system. High-speed cloud platform.

Because of this, unmanned driving is actually a completely new field for traditional car companies. Although the most important thing is still safety, traditional car companies are all taking the route of functional safety. In the field of unmanned driving, functional safety is the key far from enough.

But is unmanned driving a brand new industry? I don’t think so. When autonomous driving truly matures, it will inevitably impact traditional car companies. If users trust autonomous driving, the number of people willing to drive themselves will inevitably decrease.

So today I just want to ask you two if you are interested in participating in this car enterprise revolution with us. "

The two now understood why Meng Qian called them here at the same time. Meng Qian wanted to engage in driverless driving and didn't want to bet on a traditional car company, so there was nothing wrong with calling them together.

"When Mr. Meng is about to enter an industry, he must have already prepared for it?" Wang Chuanfu said with certainty based on his understanding of Meng Qian.

Meng Qian smiled and motioned to continue watching the PPT, "As I said just now, unmanned driving requires lidar and millimeter-wave radar. It can reach the centimeter level, and the millimeter-wave radar can make up for the shortcomings of the lidar that is not direct and fast.

Therefore, we have done a lot of research on lidar and millimeter-wave radar. For driverless driving, we have developed this 64-line lidar. Please see this dynamic picture.

The laser beam is emitted from the lenses on both sides. When encountering an obstacle, it passes through the middle lens and can be captured. The receiver can judge the obstacle through processing and analysis.

All time information, control information, and received signals will be transmitted to the top motherboard for data analysis, and the rotation recorder at the bottom can record the position and rotation information during rotation, and all raw data will be transmitted to the computer through the bottom data cable.

As for the biggest advantage of this product, it is that there are 64 beams emitted radially in the vertical direction, which can cover an angle of 26° in the vertical direction and reach a detection distance of 100-120 meters. The error of detecting the position of obstacles is about 10cm. Rotating 10 times per second, it can achieve horizontal 360° field of view coverage, and output up to 130,000 information points per revolution. "

The two people can still understand and understand this thing, and it is easier to digest, but what Meng Qian said next made them feel a little confused, "But in my opinion, our biggest advantage now is in the algorithm. .

As far as I know, several unmanned driving companies, including Google, are using deep learning for driving training, and the output of the controller is directly derived from the input of the sensor. However, as the founder of deep learning, I can basically confirm this depth. Learning is the wrong direction for self-driving cars because it's relational reasoning, not causal reasoning.

For example, people drive slowly on rainy days. The logic of this kind of deep learning is to drive slowly on rainy days, but it doesn’t know why people drive slowly on rainy days. If we don’t have data on snowy days, a situation will arise. , Unmanned driving will drive slowly in rainy days, but drive at normal speed in snowy days, and may even speed up due to other data.

This kind of algorithm without causality is a great hidden danger for driving safety, because there will be too many unknown situations in the driving process that require the computer to make causal judgments.

Therefore, we need to introduce a new decision-making mechanism in the field of autonomous driving. We have found such a mechanism, which is the Bayesian network. The Bayesian network is a probabilistic reasoning system. And event credibility analysis has a good classification effect, it has two decisive advantages: modularity and transparency.

The advantages of modularization are particularly important, such as updating the gearbox of a car. When the gearbox is replaced, we don't have to rewrite the entire transmission system. We only need to modify the subsystem modeled for the gearbox, and the rest can remain unchanged. .

As far as I know, there is only one company, Dafeng Group, that uses Bayesian networks to analyze a large amount of driving data.

And this is our great advantage. "

I don't know why, but Wang Chuanfu and Li Shufu saw Meng Qian's two leading technologies as normal, without even the slightest disturbance.

Meng Qian is actually very clear about the attitude of the two of them. Seeing that they have no intention of resisting, the matter is decided. In 2010, Dafeng Group will actively seize the high ground of human driving technology.

Thanks for the rewards on the road to Bensi! Thanks for the reward of One Yuan Taichu!