Software is a good cut for Internet companies to do autopilots.
The following content is the text version of the audio program “Observation +”. To listen to more exciting interpretations, please subscribe to “Observation +”
in the app.
Interview | 龙真梓
Documentation | Zhang Yanping
Drip has announced that it will upgrade its autonomous driving department to an independent company. In response to this news, many people think that this is the adjustment made by Didi for the listing. However, it is undeniable that in the past three years, Didi has a lot of layout in the field of autonomous driving. Not only has the laboratory been established globally, but also a new auto-driving subsidiary, Drip Woo, has been established this year. How will this adjustment affect the way the Didi will go on the market? What are the characteristics of auto-driving in traditional car companies, Internet companies and new car companies?
Observers in this issue: Luo Chao, founder of Ray Technology, well-known technology from media Luo Chao channel producers
: Recently, Didi announced that it has upgraded its autonomous driving department to an independent company. Seeing this news, many people think that Didi is following Uber’s path and stripping the autopilot section before going public to get a higher valuation.
Luo Chao: You can look at it from two angles. On the one hand, this business has indeed received a higher degree of attention. The department was upgraded to an independent company with more autonomy. On the other hand, Didi feels that autonomous driving is difficult to scale up in the short term, and if it is tied to the main company, it will drag down the performance of the main company. So splitting the autopilot part will make the main financial statements of Didi look better.
: According to the current company and business conditions, how much will the divestiture of the autopilot business play a role in the launch of Didi?
Luo Chao: The stripping of the autopilot business has not greatly promoted the launch of Didi, but this behavior can indicate that Didi is accelerating the pace of listing. The key factor in the launch of Didi is its expectations for itself. Listing is actually very easy. Last year, many companies that were not very good were listed. But until now, some companies have no market value of more than 100 million US dollars, and their performance after the listing is very poor, so the focus is on the time when Didi is listed, what kind of valuation it wants and its long-term performance after listing. These are the real concerns of Didi.
Split autopilot business For Didi, on the one hand, the current performance will be better, on the other hand, before the listing, we will not mention these uncommercial prospects. Drip and UbeThe ultimate goal of r is to operate its own autonomous fleet and provide travel services. No one buys a car in the future, nor a driver or a taxi, but to achieve this, it is still a long time later. At the pace of accelerating the listing, Didi needs to tell today’s story, not tomorrow’s story.
: In addition to upgrading the autonomous driving business, Didi has always been in the field. Since 2016, Didi has established research institutes and laboratories in China, the United States, and Canada, and has obtained auto-driving road test licenses in California and Beijing. In March this year, a new auto-driving subsidiary, Didi Wo, was newly established. It can be said that the layout is still relatively comprehensive. Now it seems that Didi has its own advantages in doing autopilot?
Luo Chao: One advantage is the application scenario. There are few application cases for unmanned vehicles, and more stay in the demo stage. For example, Google’s unmanned vehicles, it cooperates with the American network car platform Lyft to launch the unmanned car network car service. Why does Google want to work with Lyft because Lyft has a scene where users call for cars, and Didi also has this advantage. But apart from this advantage, Didi has no other advantage. First of all, technically, the distance between Didi, Google, Uber, Baidu is still very far. Because the technology of unmanned vehicles requires a lot of money and time investment, but the drops are more conservative, especially in the past year, it has focused on safety, which has become more conservative in the long-term business of unmanned vehicles. The second one is about data. Although Didi has a lot of data about the car every day, these data have no value for the unmanned car. The data that the unmanned car really needs is the road test data, and the high precision. Maps can be used as data for infrastructure.
: I said before the drop, I mainly do software in autopilot. Is it a good cut for the Internet company to make software into the field of automatic driving?
Luo Chao: It must be. Because Internet companies are best at software, they are not good at hardware. The software is also very fine. Different software has different practices. For example, Ali’s smart driving layout has an operating system called Ali OS. Its predecessor is cloud OS. Cloud OS was used to do mobile phones. The key product now is smart driving. Intelligent driving is different from automatic driving. Intelligent driving can be applied not only to unmanned vehicles, but also to intelligent driving of ordinary vehicles. Baidu also made a platform called Apollo. Apollo is not an unmanned car. Like Android, which is autonomous driving, they hope that traditional car manufacturers and new car manufacturers can use this platform to build their own cars. So Baidu provides technical services, not only provides one software, it provides veryMulti-software solution. I am not very clear about what the software is going to do.
: What are the characteristics of traditional auto companies, Internet companies and new car companies that do their own autopilots?
Luo Chao: Traditional car companies are best at building cars. They can provide good quality cars, but autonomous driving is basically based on third parties, although they have some explorations of their own, but in The software level does not have an advantage. Internet companies, especially Internet giants, are best at algorithms, especially algorithms related to AI, such as machine vision, intelligent planning and decision making, which they are good at. Unmanned vehicles also need high-precision maps, Baidu has Baidu maps, and Ali has high-tech maps. These are their advantages. New car companies like Tesla and Weilai, they seem to build cars and understand the Internet, but in fact AI technology is very weak, I think they have no advantage.
: In comparison, who do you think is more likely to have an advantage?
Luo Chao: These three roles are not horizontal contrasts, but more collaborative. For example, Ali’s cooperation with SAIC, Baidu’s Apollo also has a large number of traditional car companies, there is no strong competition between them. The core reason is that the unmanned car is also a very distant thing like the Apollo moon landing plan. Now even the market has not really formed. It is neither the blue ocean nor the Red Sea. Everyone is still on the shore. So there is no competition, and everyone is still groping. Of course, in some specific details, such as the competition for autonomous driving talents, there will still be competition for a good sensor company, but the competition is not particularly intense.
PS:If your company, industry and hot topics in the new business world are closely related and are seeking coverage, please contact us with the introduction (Contact: Long Zhenyu Contact Way: longzhenzi@36kr.com)