GNEV12 | Nezha Automobile Daniel Zhang: Smart Class Smart Driving should be used by the working class

On February 26, the 12th Global New Energy Vehicle Conference and the First Electric Vehicle Main Congress kicked off in Beijing.

There were 25 outstanding car owners representing their respective brand car owners, focusing on core topics such as endurance and replenishment, smart cabin and intelligent driving, service, community, and "Ta era", and face-to-face communication and dialogue with 25 car company representatives.

In the starry sky speech session, Daniel Zhang, co-founder and CEO of Nezha Automobile, exchanged with everyone about Nezha's intelligence and how Nezha's user experience is done.

He believes that the smart driving smart cabin must be the pain point of the user. Because now when buying a new energy vehicle, there is no large screen of more than twelve inches in the car, then this car is not a new energy vehicle. If there is no automatic driving function of level L or above, then this car is not a smart car. If it cannot continue to achieve iterative upgrading and can grow on its own, then it cannot be defined as a new energy vehicle. Therefore, consumers, especially young consumers, have a lot of demand for smart class intelligent driving.

Daniel Zhang said that Nezha Automobile wants to achieve a special performance car, the price can rise again, but our original intention has not changed, ordinary wage earners, buy fuel cars are Volkswagen Toyota, or even BMW, I think as long as it does not exceed 300,000, it is to build cars for the people. Of course, we will continue to be within 200,000, 100,000 cars, continue to have, we will build all the cars as smart cars.

Daniel Zhang, co-founder and CEO of Nezha Automobile

The following is a transcript of the speech (slightly abridged):

Daniel Zhang: Good afternoon, distinguished friends! It is a great honor to come to the first electric vehicle owners' congress of Pang's first electric network, which should be the twelfth annual event held by Pang. I am afraid that I have been participating since the first few sessions, the original theme may not be this, the original theme is based on enterprises, manufacturers, this year is to do an innovation, is to the user main. So today, on behalf of the United Congregation, I am here to communicate with you.

What I said today is a propositional essay, which is completely a box that Teacher Pang has given me to define, and let me speak in this scope. So here's an opportunity to talk to you about our intelligence and how our user experience is done.

The first question, is the smart driving smart cabin a pain point for users?

Without a doubt, it must be! Because now when buying a new energy vehicle, there is no large screen of more than twelve inches in the car, then this car is not a new energy vehicle. If this car does not have an L-level automatic driving function, then this car is not a smart car. If this car cannot continue to achieve iterative upgrading and can grow on its own, it cannot be defined as a new energy vehicle. Therefore, consumers, especially young consumers, have a lot of demand for smart class intelligent driving. Of course, we also know that just now I also listened to the speeches of some user friends, users are still looking forward to the products we currently provide, but there may be some problems.

The first expectation is some early technology, which is quite a lot of ideas. Can there be some better, fresher features for everyone to try out? This is how to balance the balance between technology leadership and user needs, how to make your products meet the needs of users, and tap the real needs of users. I'm referring to the real need, what are its pain points? There are some fresh features that can be used.

The second understands me better and is smarter. I know that there are now some smart driving systems that should not exit when they should quit, and they cannot quit in time when they should quit. Some voice systems are more silly, navigating the large-screen system of the car, and the navigation function may not be so perfect. Especially in the location and use of charging piles, it is not so smart and the practicality is not so high.

The third issue is security. In fact, if smart drivers want to achieve L4 or even L5 fully autonomous driving, I think there is still a long way to go, the long tail effect in various scenarios is too long, and there are many situations that cannot be exhausted. Especially now that this kind of automatic driving at high speed is estimated to be the first to be achieved, it is a closed scene. At low speeds or in parking lots, it is also possible to partially use autonomous driving. But the complex roads of the city, the last five kilometers to say fifty kilometers, this is very difficult to achieve. It may take a considerable period of time, and more vehicle-to-road coordination may be required. Then in this case, safety is the primary issue, the state is also through various regulations, various laws and regulations of the provisions of constraints, including the management of the data platform, can make the vehicle itself become safe. More importantly, in the future, your car can not get out of control, can not be used by some bad people, etc., the safety of the car, the safety of driving, the safety of data, the safety of pedestrians, this is a topic that needs to be solved.

Then there is also an iterative upgrade. Can we buy back a car like a computer, a mobile phone, can be constantly iterative upgrades, can continue to have new applications, new experiences come in, I think this is the user's four aspects of expectations.

Now in fact, there is still a gap between the user's expectations and the products we provide now.

I think the first is that the operation is not smooth, and the response of the interaction is slow. Many of these problems are not only software problems, application problems, many times hardware problems. Because there are more and more screens and more and more signal inputs, if your hardware platform, the electronic and electrical architecture is not enough to support the operation of such a large amount of data, which is actually very problematic. Therefore, in order to solve the problem of smart cockpit and driving, we must first solve the problem of price.

The second problem is that the ecology is not rich and the content is relatively single. In the car, everyone thinks that the car will be a third space in the future, and it can do everything in the car. But now in fact, there is still a gap between the car and the real third space, and we are still mainly doing it as a driving means of transportation. It is nothing more than adding some entertainment elements to it, but there is still a gap between the real office, home, rest, and conference management. The content of the app is not rich enough.

The third travel scenario, I mainly refer to automatic driving here. We now have a lot of scenes that can't be covered, now for example, we do an application, vehicle summoning, automatic parking. Of course, many times you have concerns, so what happens when a child or a dog comes out of the car next to you? Then when you go out of the neighborhood, there is a tricycle next to you, and a cyclist comes to say goodbye to you, what do you do? Many scenes cannot be covered. When you use it, you will feel the frequent introduction of vehicles, even on normal road driving, there is a phenomenon of frequent introduction. So this is the insufficient coverage of the application scenario.

The last intelligent experience is less, and there are fewer personalized customization services. What vehicle companies are doing now is to introduce charging piles. But the charging pile is not in use, the charging pile is not broken, in fact, the information here, a large amount of information has not been imported, or it needs to be continuously improved. Including our Nezha enterprises, not to mention other application services. We should divide the needs of users into basic services, value-added services, and some such services that exceed expectations, little by little, a project, a project to do it well.

The trend of the industry, I think there are now these four aspects, the first is data-driven, we know a solidified operating system or a smart cockpit system, relying on the back-end manpower to continuously iterative upgrade, you can't do it! Only by constantly relying on neural networks to learn, data to keep improving, build a model, let the machine, let the device, let the operating system learn from my drive. I think this ability is the core competitiveness of future car companies. For example, road information, because the road information changes every day, then your now everyone says that people and machines are driving together, high-definition maps are introduced, and the map changes the condition of your car. In a very small and small segment of the scene, your data needs to be updated and your strategy adjusted. For example, in the future, is it possible to have a car, you are a female user, then you like to drive is gentle, like to change lanes when it is faster, a variety of learning abilities are required.

So this requires a lot of data to drive. Big data platform, the construction of big data center, this is the next step that all car companies should make efforts.

The second is the increase in computing power. Because there are more and more screens and more data, this system needs huge computing power to operate, including some redundancy, especially the redundant system can not have only one, it has several sets of systems. So the demand for computing power is getting bigger and bigger, so we see that there are the original eight sets, thirty-two sets, sixty-four sets, one hundred and twenty-eight, two hundred and so on. It is said that there are now two thousand and one thousand, which is on the one hand, and the computing power needs to be improved. But we really don't forget that cars are costly, all the stacking of computing power is the need for users to pay or enterprises to pay, so the competition of computing power can not engage in an arms race, the first user to bear this cost, we are now Nezha S, two lidar, with high-definition maps, this system is close to thirty thousand. I think this two or three years, three or four years is completely enough.

We are communicating with friends, including companies that are relatively advanced, in fact, in three to five years, this system is enough! Absolutely capable of covering most application scenarios. Two hundred or five hundred computing power is enough, more is waste, more is proof that your enterprise algorithm is too poor, your operating system or your application layer is written too badly, your code is too poor.

So one is that enterprises should improve their own algorithms, their own software, and their own capabilities to become more and more refined, and the second is to reduce the cost of users. Therefore, there is no arms race. Just now I saw that Mr. Ding also put forward this point of view, I think it is very good! It cannot be said that the higher the computing power of anyone, the greater the number of people who feel that they are strong. Who can use the most users at a low cost, the smartest feature is called a cow.

The third is the upgrade of the architecture, which I just said, in fact, all kinds of application systems are like a highway. Self-driving, cockpit, entertainment, navigation, we see the same car, a car in all directions. But the car is doing more and more advanced, but the highway is doing very badly, can't run, or even has the risk of overturning, it is very finished! So the new battery architecture, from distributed to regionally controlled, and finally to central group units, is a path of future evolution. Of course, the central group unit is not the only one, there must also be provincial roads, national roads, country lanes, capillaries.

The last lidar, so look at it now, this year's lidar is most likely to be equipped with some advanced models, because lidar does solve the problem of recognition difficulties under many scenarios, too far, the road conditions are not good, and its recognition sensitivity is better. In order to make their cars smarter and have more application scenarios to use, many models of lidar will be equipped, and I think this should be a big trend. Because that's what we do.

So what we did, here I will briefly report to you, the first one is that we are also saturated in software algorithms, we do not do hardware or temporarily do not do hardware. But in the software algorithm, the operating system we invested a lot of manpower and material resources, I think this team in all the car companies should be considered to be relatively advanced. Our investment in intelligent technology is no less than that of other enterprises, whether it is people, resources, or funds.

As I said just now, it is the first architecture, we are now the intelligent driving domain, the body domain, the power domain, the cockpit domain four domains, the next step to do the central computing unit, really have a smart brain to control the subsystems, this is what we are doing. We hope to build our provincial highways and national highways, and then we will do a better job in domain control, in the cockpit.

The second is voice. The future interactive system has a lot of innovation, I think in a few years, the real sense of the screen is not, there may be screens everywhere, as long as you think it is a screen it may be a screen, so the future delivery method will have some big subversive changes, and even some wireless connections on the whole car will be a trend. It is it that leads to this change in the form and the way the entire vehicle is delivered. What we call the chassis now is actually to achieve the automatic driving function of L4 or above.

Because the traditional mechanical, hydraulic control is not to reach the sensitivity of milliseconds, you want to make the car smarter, all kinds of situations can cope, must be a wire control, must be a data drive rather than rely on mechanical. So man-machine co-driving, how to integrate the current human and machine driving vehicles, when it is time to launch, when it is time to access access. For a long time, whoever wants to make this piece smoother can win.

The last ecological construction. To continuously introduce more service content, the original may be a radio to make a call, and then add navigation, now plus charging services, as well as play games, sing karaoke, in the conference system and so on, a lot of services will be added. This kind of ecological construction and service content construction are the key tasks we will do below. That's the complex scenario we're doing, we're dividing it into high school scenes, urban, the last five kilometers, the last kilometer of low-speed enclosed parking lots, we divide it into four scenes, and we're going all out.

I think with lidar, with high-precision cameras, plus sensors, a chip with high computing power, put our algorithms, make our systems smarter. We look forward to seeing all of this capability achievable at 200tops. That's what we should be doing now. We're also doing some teams of a thousand people, doing some self-developed announcements, which is algorithmic, software, data platform, we all have to do some innovative work. Including the underlying software, some efforts are also being made.

In order to achieve a particularly good performance car, the price can rise again, but our original intention has not changed, ordinary wage earners, buy fuel vehicles is Volkswagen Toyota, or even BMW, I think as long as it does not exceed 300,000, it is to build cars for the people, of course, we will continue to be within 200,000, 100,000 cars, continue to have, we will all the cars as smart cars to build. L2-level autonomous driving functions can also be achieved within 100,000, which is what we are doing. We want to build a good platform and then keep doing some cropping. We have realized the Nezha V, 70,000 or 80,000 cars, we can achieve L2 level intelligence. Unfortunately, because of the chip problem, the delivery is now slower.