The platform has chosen to combine with the car company.

BYD played the "new energy leader" LOGO for the new forces around him; Li Bin, chairman of Wei Lai Automobile, communicated with various people who were concerned about his car building project in the morning. His face was only excited and not tired. On April 25th, the Beijing Tianzhu E3 Museum almost became a charter for new forces to build cars. The main host factories, public relations, and media people who travel in the East and West directions of the pavilion can clearly feel the difference between the atmosphere and other venues.

In just two years, with the start of the scale of China's new car building forces, the product network of the auto show has begun to connect with the traditional car building thinking. At the same time, the on-site model verification of autopilot in Beijing's Tianzhu venue is also accelerating.

Define "Chinese-style autopilot"

On April 23, Gujunli, a former Tesla Autopilot core R&D member, joined Xiaopeng Automobile as vice president of auto-driving R&D. For the first time in Beijing, he spoke: Automated driving systems that adapt to China's road conditions need to detect obstacles and traffic environment information that are more than three times higher than in the US and Europe. ”

The tough female doctor wrote 50 % of the code for the Tesla Autopilot 2.0 core unit in a matter of months. In her view, the lack of accumulation of actual driving scene data, many self-driving functions in China "acclimatized" or even "nothing", can not adapt to the actual driving conditions, traffic density and driving habits of Chinese users.

In the face of specific scenes, Gujunli gave an example of parking to the "Huaxia Times" reporter: It is also a side parking. Because of the initial learning ability of autopilot, it will repeat the experience it had acquired the previous time. If you're in a parking environment where parking is regular, it's regular, and if everyone doesn't behave, the second and third automatic parking starts to behave.

This raises the question of how bad car environments can make it harder to judge machines on the one hand, and the learning and cognitive abilities of machines on the other.

As a Super product manager, Hexiaopeng, chairman of Xiaopeng Automobile, immediately saw the nature of the problem, that is, almost all of the current automatic parking can solve the actual use scenario may not exceed 5 %. When the resolution rate of a user scene does not exceed 10 %, it means that the user's use frequency and usage inertia will be very low.

The direction of the Super product manager determines the direction of scientists.

For this reason, the team led by Gujunli insisted from the beginning that the inevitable trend of autopilot evolution is the strong localization of software, and that a set of systems and systems can never be universally applicable.

In order to complete the data accumulation to provide the underlying foundation support for the upgrade iteration of autopilot technology, Xiaopeng Automobile named its own self-driving system "XPilot". At the same time, hundreds of Xiaopeng 1.0, which was mass-produced at the Haima Zhengzhou factory last year, have all been distributed to Xiaopeng employees and enthusiasts across the country. The cumulative mileage has been nearly 5 million kilometers, and Xiaopeng has contributed to the background every day. A lot of data.

Platform selection and car company integration

In an interview with the "Huaxia Times" reporter, Hexiaopeng said: "Considering the need for Chinese-style autopilot routes, this has become a realistic consideration that must be faced before the product's quantitative scale is delivered. In 2018, he expects delivery of products where the big new forces have made real breakthroughs in self-driving technology to be delayed.

In his view, in the field of core differences such as autopilot and smart network connection, it is possible to make high-intensity investment at a certain point and invest in a single point breakthrough in large-scale superior resources; However, on the road to large-scale and quality delivery, slow is fast, so Xiaopeng gave up the familiar Internet product building method, did not choose a quick trial and error, rapid iteration of the product concept, but slowly polished the product, Little by little improve the user experience.

This process from the technology dimension to the pre-architecture, to the bottleneck breakthrough, and to the delivery of finished products, let Hexiaopeng, who was born on the Internet, deeply realize that there is a need to respect the traditional industry.

At the same time, Baidu is in alliance with BYD, Tencent is in alliance with Chang 'an, and smart driving technology must find realistic manufacturing carriers to provide hardware and software support during the platform opening process.

On April 19th, Baidu Apollo, the world's largest auto-driving ecology, announced its 100th strategic partner, BYD, on its first birthday party.

Why did Baidu choose to cooperate with BYD? Earlier this year, Mr. Wang announced a major strategic shift by BYD to open its supply chain to auto companies around the world, as well as the opening of the bottom layer of its hardware platform to software developers around the world to promote applications based on car hardware.

Baidu is clearly aware of the value of an open platform for car hardware. For the most important actuators of autopilot, wire-control movements, and even the best suppliers in the world, there are still a lot of problems with software control strategies.

On the evening of April 23, Zhangbaolin, chairman of Chang 'an Group, once again mentioned the strategic cooperation between Chang' an and Tencent. The old car man even expressed the need for this cooperation in terms such as "the era has eliminated you and has nothing to do with you." Ma Huateng recently signed a strategic cooperation with the car company in succession, revealing Tencent's concern about the vehicle and the underlying control business of the vehicle.

The Deep Meaning of Re-travel

While Xiaopeng, Baidu, and Tencent join forces with traditional car companies to build self-driving bottom business, the travel industry is also lurking.

On the morning of April 24, Drip announced that it would jointly build a consortium of 31 automotive industry chain companies, established a flood league battle, and build a vehicle operator platform. Drip will open online car contracts, time-sharing leases, custom cars, smart driving and post-market service businesses. Drip also released D1, a shared travel intelligence tool designed and produced with the company.

Liu Qing, president of Drip Travel, invited Xu Liuping, Chairman of FAW Group, Xu Heyi, Chairman of BAIC Group, Wang Chuanfu, Chairman and President of BYD, Feng Xingya, general manager of GAC, and Li Xiang, founder of Car and home. The youngest of them, Li xiang, said that the car will be self-driving in the next 10 years. In this space, passengers can work, be entertained, socialize, and have meetings, which is almost the same as their company’s name. Car and home.

Drip 's founder, chairman and CEO, Chengwei, said Drip wants to be the world's leader in smart transport technology, investing in smart transportation and driverless. Chengwei said: "100 years ago, the car was invented by Mercedes-Benz. We hope that the first generation of shared travel networks for shared design and large-scale operation will be born in China. "

Liu Qing and Li Chengwei accelerated their layout in the international market. There is a template called Lyft.

In fact, in early 2017, the company still seemed content to work with a large number of auto makers, simply deploying their cars to their own pick-up platforms, including Waymo, General Motors, Jaguar Land Rover and Ford. But by mid-2017, Lyft announced that it will develop its own autopilot system and intends to provide it to other manufacturers.

Recently, Magna Group, a global supplier of auto parts and components, invested $200 million in Lyft and will work with the car service company to develop highly self-driving cars. In addition to automation software, Magna and Lyft's alliance covers almost every aspect of autonomous car development.

Drip’s vision for the future includes two figures, one for the two billion users it wants to serve, and the other for the 50 % of travel that a shared travel network can carry.

In just one month, a Chinese-style alliance of data based on self-driving technology set sail from Beijing, followed by the Chinese core built by new car building forces, BAT, travel platforms, and traditional car companies. The Chinese model of self-driving continues to be enabled.

The page contains the contents of the machine translation.