How about the autopilot technology of Xpeng Motors? Will it converge to Tesla?
How about Xpeng car? Sales is the best answer. A few days ago, Xpeng Automobile announced its delivery performance in September. The data shows that the total delivery volume of Xpeng vehicles in September reached 3,478 vehicles, up by 145% year-on-year, setting a new monthly delivery volume in 2020; Tucki P7 delivered 2,573 vehicles in a single month, setting a monthly delivery record. Judging from the current production capacity of Tucki P7, this achievement is very worthy of recognition.
At the same time, Xpeng Motors said that more than 93% of all delivered vehicles chose the smart version equipped with XPILOT intelligent driver assistance system. In other words, most consumers are very interested in the intelligent driver assistance system of Xpeng Motors, and it can even be said that this system is the decisive factor that affects these users' decision to buy Xpeng Motors. Then let's talk about the intelligent driver assistance system of Xpeng Motors.
First of all, we need to put the time back to 20 19 and look at the feud between Xpeng Motors and Tesla. In March last year, Tesla sued its former engineer Cao (? Cao (transliteration of Chinese name) was accused of stealing trade secrets related to the autonomous driving system and taking it to Xpeng Motors.
Cao Guangzhi, a key figure in the lawsuit, was a computer vision scientist at Tesla. According to Tesla, he worked on the development of autonomous driving system and was one of 40 people in the autonomous driving system team who had access to the source code. However, on October 3, 20 19/kloc-0, 65438+ Cao suddenly submitted his resignation to Tesla, and then moved to Xpeng Motors as the perception supervisor.
Tesla said that from 20 18 1 1 to 20 19 1, Cao backed up Tesla's entire repository, autopilot and neural network source code library, and uploaded them to his personal iCloud account, totaling more than 300,000 files and directories related to the autopilot system. Tesla believes that Zhihui and Xpeng Motors have no legal right to use the relevant technologies of Autopilot system, which is the result of Tesla's research and development and team efforts that took more than five years and hundreds of millions of dollars. However, at that time, Xpeng Motors responded that before and after Cao joined the company, the company did not find what Tesla called violations.
However, in the following July, Cao Guangzhi admitted in the documents submitted to the court that he did download Tesla's technical documents, that is, uploaded files containing relevant source codes to his personal iCloud account, but denied that any trade secrets related to Tesla's autonomous driving were transferred to Tucki. At this point, whether Cao provided Tesla source code data with its backup has become the focus of controversy in this case.
On April 25th this year, Xpeng Motors issued a statement saying that after Tesla sued Cao, the employee who moved, it was an "unreasonable request" to release all the source codes of autonomous driving. After all, in the past year, Xpeng Motors has taken the initiative to provide an electronic backup of Zhi's work computer, and also allowed Tesla to contact the company's source code library for evidence collection according to the court protection order before the prosecution date. But obviously, Tesla didn't find the key evidence from it, so he asked to expand the investigation.
In fact, not only in the field of autonomous driving, but also in many technology-intensive industries, there will also be various disputes over trade secret theft caused by "talent flow". Even if there is no dispute, when a company's technical core personnel quit to another company of the same type, it will also cause various speculations and concerns in the industry. Just like Gu, former vice president of autopilot research and development at Xpeng Motors.
It is understood that Gu Lijun is a famous machine learning expert and doctor in Silicon Valley, USA. 2065 438+02-2065 438+06 Chief Engineer of international chip giant AMD, responsible for the research and development of big data and deep learning software; And in February of 20 16, he joined Tesla, formed a Tesla machine learning team and developed Autopilot? 2.0 products and large-scale deployment on multi-generation Tesla cars; 2065438+07171On October 23rd, Xpeng Automobile announced that Dr. Gu, a famous machine learning expert, was appointed as the vice president of Xpeng Automobile Autopilot, reporting directly to the chairman of the automobile. But in March of this year, Gu left her job. The successor is Wu Xinzhou, the former head of Qualcomm Autopilot R&D team.
It can be seen that in the past two years, there have been more and more personnel flows between the Silicon Valley autonomous driving giants represented by Tesla, Apple and Qualcomm and the new car-making forces represented by Weilai and Tucki. So apart from the close relationship between Xpeng Motors and Tesla, what level has it developed in the intelligent driver assistance system? Take the production model P7, which best represents its technical strength, as an example, we can roughly understand this situation.
Autopilot requires ASIL-D hardware SoC, while NVIDIA's Xavier and Orin can reach ISO? 26262? Safety standard of ASIL D system. No matter Tesla, Tucki or Ideal, they chose NVIDIA for the first time to prepare for independent research and development of autonomous driving. It can be seen that NVIDIA's autopilot chip must have many advantages.
XPILOT carried by Tucki P7? 3.0 automatic driver assistance system, the main chip is NVIDIA? Driving? Xavier, with L4 autopilot computing capability, single-chip computing capability of 30TOPs and power consumption of 30w. Although the computing power is not as good as Tesla FSD? The performance of 70TOPs is 12 times that of the current mainstream EyeQ4 platform, which is enough to support the demand of L3 mass production and future L4 level testing.
In order to ensure the normal operation of the system, Tucki P7 matched 14 camera and 5 millimeter-wave radars, 12? Ultrasonic sensors constitute the only 360-degree dual-sensor fusion system in the industry. Among them, the far, middle and near three-eye cameras and the side cameras cover more than 180 degrees, which can realize lane line identification in narrow corners and short-distance traffic jam identification; In the parking lot scene, two side-looking cameras can identify the side-looking parking space, and realize automatic parking for the first time in the whole industry without crossing the parking space.
On the positioning level, Tucki P7 uses high-precision maps with decimeter precision, which is the basis for realizing lane-level positioning; At the same time, the use of triple high-precision positioning hardware (GPS+RTK+IMU) makes the global positioning accuracy reach centimeter level; In addition, the relative positioning accuracy of P7 is less than 0.3% due to the use of instant positioning and map construction technology (SLAM). The three technologies complement each other and overlap each other, which can effectively improve the stability and safety of autonomous driving in overpasses, tunnels or bad weather.
The technical strength of Xpeng Motors is closely related to its emphasis on R&D investment. According to the data in the prospectus, as early as 20 19, Xpeng Motors invested 2.07 billion yuan to develop autonomous driving and intelligent operating systems, and the R&D expenditure accounted for 89.2% of the company's total revenue, which was about double the investment of 105 billion yuan in 20 105, and it was also one of the enterprises that paid the most attention to R&D investment in the new car-making army.
In the aspect of automatic driving data collection, Xpeng Motors collected about 2.9 million annotated pictures (such as road character interference, rainstorm, night, occlusion, etc.). ), and also deeply studied and studied the neural network technology, and developed the functions of natural language processing and natural language understanding.
In addition, at the end of last month, Xpeng Motors also announced that it had reached an agreement with Guangzhou Kaide Investment Holdings Co., Ltd., a wholly-owned enterprise of Guangzhou Development Zone Management Committee. Kaide provided 4 billion yuan of financing to support Xpeng Motors to accelerate its business expansion, including building a new intellectual base and purchasing factory equipment in Guangzhou Economic and Technological Development Zone. This will also become the second self-built factory of Xpeng Automobile after Zhaoqing Factory.
On September 24th, the first batch of Xpeng cars 100? Tucki G3i was also officially sent to Norway, which was the first order for Xpeng in the European market. Norway 165438+ 10 in 2020. Norway has always been one of the countries with the highest acceptance of electric vehicles in Europe. In August this year, its market share of electric vehicles has reached 70.2%, higher than last year's 49%. BEV models account for 53% of the automobile market. As of August, the annual sales volume of electric vehicles was 54,000. If Tucki G3i can gain a firm foothold in the Norwegian market, its globalization process will be guaranteed.
At present, in the domestic market, regardless of its connection with Tesla or the comprehensive strength of Tucki P7, it is the most direct competitor of Tesla Model3. Compared with previous years, Tesla's products are completely one or two generations ahead of competitors at the same level in the market, but now car companies are more and more similar in the research and development, upgrade and improvement of driver assistance systems, which may also be an important stage performance of the rapid development of technology.
Figure? |? From the network
This article comes from car home, the author of the car manufacturer, and does not represent car home's position.