Smart car, why is so close to you

Over the past 100 years, the traditional automobile industry has finally ushered in great changes. This car-building movement from the AI (artificial intelligence) proposition is setting off a new round of upsurge.

In order to take the lead in this giant industry, traditional car manufacturers and Internet car companies have begun to exert their efforts and layout, and the car building movement is in full swing. "Automobile is an existing very huge industry, with an international market size of 35 trillion and a domestic market of up to 10 trillion. Artificial intelligence technology, for the automobile industry, is an unprecedented subversion, and it is also a rare opportunity to make more Many companies can enter and participate." said Zhu Yan, a partner of Weilai Capital.

However, it is not easy for more car companies to make the concept of smart cars (including autonomous driving and driverless cars) practical. Today, since the car-building movement blew the horn, many companies have died down and left the scene. Behind this seemingly opportunity, there are hurdles in front of car companies. Only by breaking through these many barriers can companies finally see the light of day.

Self-driving cars cost too much money, and you have enough energy?

The origin of all car-building problems is inextricably linked with funding. It is a relatively long distance from concept to implementation, from implementation to mass production, and then from mass production to sales. And every step in between requires financial support.

Apple, which has a cash reserve of 230 billion US dollars, can be described as strong. However, after vowing to enter the field of unmanned vehicles, it began to retreat steadily in the research and development stage, and finally had to lower its focus on unmanned vehicles. In anticipation of the announcement, the car-building plan failed.

In contrast, LeEco's car-building plan is a bit of a "gamble on tomorrow". LeEco Auto consumes tens of billions of RMB. After the loophole of LeEco's unmanned car cannot be covered by the capital, the once-fashioned car-building plan has also begun to be suspended for an unknown time limit.

Zhu Yan, a partner of NIO Capital, said: "If you want to make a complete vehicle, the investment is indeed huge, so for many new car companies, it is necessary to focus on the manufacture of automobiles and the production of complete vehicles. Difficulty is awe-inspiring. It is very difficult to truly deliver a car from R&D to final delivery to consumers without the investment of billions, tens of billions of RMB or even more.”

From the purchase of accessories and product development in the early stage, to the brand building and after-sales maintenance in the later stage, there is no need to consume a lot of money. This is a money-burning game. Without funds, it is impossible to go to the second half of the car-building plan firmly.

First of all, hardware cost is one of the factors restricting the mass production of smart cars. The lidar used by Baidu alone costs as much as 500,000 yuan, while the Velodyne radar used by Google in its prototype self-driving car costs as much as $70,000.

"At the level of autonomous driving, the more important hardware components are sensors. Sensors like lidar are still very expensive. The price of such a sensor basically reaches the level of a luxury car. In terms of production, there is indeed a lot of resistance." Zhu Yan said.

In terms of R&D costs, the investment in smart cars is also very high. Just looking at the salaries of driverless engineers, it's staggeringly high. According to Google's public information, in the past few years, they paid a total of 120 million US dollars in salary to Levandowski, one of the leaders of the original self-driving car project. Similarly, domestic car companies are not low on asking prices. The self-driving start-up company founded after the departure of two technical backbones of Baidu's self-driving department has a valuation of more than 100 million US dollars in the PPT stage alone; let's look at Weilai Automobile, which has a R&D team of about 2,000 people, although it has not been announced. Research and development costs, but judging from its practice of poaching relevant talents from Jaguar Land Rover, its investment will only be high or low.

Burning money is not terrible, the output ratio is the key. The current problem is precisely the mismatch between expectations between high costs and expected sales. Judging from the brands that already have quasi-mass-produced cars, it is one of the biggest problems for car companies to achieve mass production whether smart cars can still achieve high cost performance after considering market acceptance and return on investment. From the current point of view, after the injection of heavy capital, it will be difficult to achieve a high return on the cost of mass production vehicles, which has also caused widespread controversy and discussion in the industry.

However, even at this point, funding concerns haven't stopped abruptly. Car companies build cars. As a new brand, smart cars need to be gradually recognized by the market, and investment in building brand influence in the later stage is essential. Maintenance and after-sales service will also be another high expense for car companies.

In Tesla's previous earnings report, there is a fund trend that is quite interesting. Some people pointed out that this is Tesla spending a lot of money on maintenance service problems and corresponding delays, such as solving driverless driving. Safety problems caused by the end-to-end feedback, and vehicle driving problems caused by slow end-to-end feedback, etc.

It can be seen that the financial challenge is a constant for car companies. The high cost pressure is also an inevitable process for the initial development of an industry. Only after gradually accumulating and forming a scale effect can we finally keep the clouds open and see the moon. Of course, car companies form alliances and form a mutual aid model with each other, which will help this day come earlier.

"We are focusing on the field of connected car accessories, new sensor solutions and companies that can significantly reduce costs. In fact, if this sensor can be made with semiconductors, the cost will not be so high. Because the semiconductor industry will quickly reduce costs, there will be a strong scale effect. After the cost is reduced, it is very possible to achieve mass production." Zhu Yan said that he is full of confidence in the future of smart cars.

How long does it take from design to implementation?

This is a new challenge, as designers need to embrace the first big change in the auto industry in more than 100 years -- making self-driving cars the primary function of a car.

Design is a critical process in automotive manufacturing. Taking Toyota as an example, the design standard for a non-major spare part is a 1,000-page book, which shows how complicated it is. In the face of smart cars, what makes the challenge escalate is that designers must open their minds again, subvert the previous inherent cognition, and let autonomous driving technology reshape their design thinking and mode. "The challenge comes from having so many technologies - information, entertainment, connected cars - we need to be able to control them," said Ian Callum, Jaguar Land Rover's design director.

Thomas Ingenlath, Volvo's vice president of design, also said that now he needs to spend more time considering the layout and screen size of electronic equipment. As a designer, Ingrat now spends a third of his time researching new technologies with people in R&D, digital, and electronics, and discussing how these functions can be presented.

"For example, at the beginning of the design, the designer should consider the function of autonomous driving and reserve the position of the sensor; how to reduce the space occupied by the motor and expand the space for the occupants in the car; or, even if only how to These designers have to consider installing the location of the car sensors more concealed. With the improvement and perfection of technology and functions, car companies also need to upgrade the existing cars.” Zhu Yan gave an example.

The difficulty is not only reflected in the design of smart cars, but also in the realization of its technology. The design cycle of traditional car companies' fuel vehicles is long, and it generally takes 5 or 6 years from design to market. A complete vehicle needs to go through thousands of processes from design and development to mass production and delivery.

"Designing a car is very different from building a car. It's better for car companies to find a good OEM to help manufacture." Zhu Yan said. For car companies, it is more appropriate to directly use the mature achievements of the industry and find suitable suppliers for cooperation. However, a new problem has arisen. The complexity of smart car technology and the characteristics of low demand often make suppliers shy away. How to win the favor of high-quality suppliers is a headache for many car companies.

From the perspective of car companies that have successfully built cars, the process verification and testing work after the release of models also requires a certain period of time. For mass-produced cars, the most important thing is to be able to withstand various tests and have full safety and reliability. Therefore, finding a site for testing, testing in different environments, etc., requires a set of well-established plans and processes to achieve. It is understood that it took Audi A7 8 months to test the reliability of autonomous driving technology.

From this point of view, car companies need to think ahead and make arrangements in advance in this regard. "Actually, as long as you design well and the supply chain system is well managed, companies with a certain level of skill in general can produce high-quality cars. Moreover, from the time of car design, the final consumer's satisfaction level has been determined. and sales." Zhu Yan once said in an interview.

Is it difficult to overcome technical difficulties?

From Newton's universal gravitation to China's four great inventions, it is no exaggeration to say that any technological innovation or change will cause or bring some controversy. If a new technology wants to take root in the original social soil, it must accept the test of the environment.

The development of smart cars is going through such a challenging phase. To a certain extent, a smart car is a robotic system with three aspects: control, perception and path planning. Gao Hua, assistant to the president of Singularity Automobile, said in an interview with "CEConline" that among the control systems in smart cars, the most important and the most concerned is "intelligent network connection", but because the technical threshold of intelligent network connection is very high At the same time, it involves the development of technologies similar to autonomous driving, so domestic car companies will face challenges from many aspects, such as the restrictions of policies and regulations, the limitations of testing conditions, and the scarcity of high-level technical talents.

In terms of perception systems, domestic car companies are faced with unavoidable technical barriers - the core sensor technology and related products are almost all from abroad, and it is difficult for Chinese car companies to master the most cutting-edge and core technologies. In addition, in terms of environmental perception technology, China's smart cars are currently unable to adapt to complex driving scenarios and cannot effectively perceive the driving environment, which is far from the technology mastered by foreign car companies and research institutions. What car companies should also consider is whether the introduced sensors can adapt to the navigation systems of all road conditions and various environments in China after they are fused and integrated into key car functions.

Stanley Ng, CEO of MapKing International Limited, shared the difficulties in the path planning system. When the path planning of smart cars involves local paths, it tests whether the system has The ability to process information in real time, and how to quickly adjust the driving route once an emergency is detected, these are the difficulties that the route planning system needs to optimize at present.

The immaturity of the above smart cars in the technical field has transformed into concerns about the cruising range and safety performance of smart cars for non-professional consumers. If car companies want to fundamentally solve the problem, they can only focus on improving the technical level and optimizing manufacturing materials. From the perspective of the development of new energy vehicles around the world, due to the rapid technological update, consumers' acceptance of product upgrading is still limited, and it is difficult for consumers to change their user thinking in a short period of time, and they do not really accept driverless driving in concept. and other high-tech technologies in the application of automotive terminals.

"Mature and mass production will eventually be the general trend in the development of smart cars." Gao Hua pointed out that for companies with technology and Internet backgrounds, they have the advantages of technology and talents. Data services will make a difference. As a traditional car factory, it can cooperate with technology companies in terms of technology and talents to implement intelligent networked vehicles more quickly. At the same time, the development of the industry will also spawn a large number of third-party technical service companies, focusing on the development of more advanced sensors, higher performance and lower cost and power consumption chip platforms, and even like Udacity, it can help the industry to quickly cultivate artificial intelligence direction. talent.

It is imminent for the industry to establish a core technology alliance. The reason is that the scope and field of smart cars are very wide. In the Internet of Vehicles and other aspects, companies need to cooperate with many peers. The formulation of national and industry standards also requires Combined with the technical capabilities, application levels, market demand and other actual conditions of existing enterprises in the industry. "Smart cars will eventually enter the homes of ordinary people, bringing huge changes to our future transportation and travel experience and industry ecology." Gao Hua said.

Can't you have both hands-free and safe driving?

At the 2017 Baidu AI Developers Conference held not long ago, the incident of Robin Li, founder, chairman and CEO of Baidu, attracted widespread attention. In the video, Li Yanhong rides in the co-pilot position of a driverless car. On the 15-kilometer road, the driverless car drives smoothly. However, Chinese law makes it illegal for drivers to take their hands off the steering wheel. It is understood that China's "Road Traffic Safety Law" has not been able to take into account the issue of driverless vehicles, and the same is true of the relevant supporting legal system.

The safety of smart cars is mentioned again. The original intention of autonomous driving is to improve the traffic safety rate, but before its popularization, driving safety is the primary goal. Different architectural plans in cities and towns make road conditions very different. How to make vehicles suitable for autonomous driving under different conditions? How to ensure the safety of the car when driving autonomously? Behind the intelligent technology, a large amount of behavior analysis and learning of people is required, and "sensor fusion" and big data processing are also required.

Some companies are already testing this, such as Cadillac's Super Cruise, which uses cameras to ensure drivers are fully prepared to take over the vehicle; IBM researchers recently filed a patent for an AI system that It can determine whether it is safer for a human or a system to control the vehicle in any scenario; Tesla also uses a series of sensors to help drivers monitor Autopilot conditions.

In addition, the route planning system can also play a role in avoiding driving hazards. Guotu Consulting Co., Ltd. can provide relatively powerful information capacity, local voice driving guidance and 3D operation interface to obtain information for drivers in advance and plan a safe driving path. However, the above functions need to rely on a strong 4G or even 5G network. Can accurately send SMS coordinates, weather disaster information, etc.

As for the popularization of autonomous driving technology, Tian Feng, executive vice president of ZTE Intelligent Vehicles, also shared his views, "Intelligence and networking will have a stronger impact on the traditional automobile industry than many people think. Stronger and faster." The auto market predicts that autonomous driving technology will strike in 2030, but in fact, autonomous driving technology has matured, and what needs to be changed is the acceptance and support of new technologies by consumers' perceptions and social environment.

However, the current laws and regulations cannot keep up with the pace of technological development, and the lack of laws for autonomous driving is a problem all over the world. The technology in the field of smart cars is changing very fast, and relevant standards and regulations must be able to follow the pace of technology updates in real time to escort the landing and application of high-tech concepts. At present, the Chinese government has also carried out legal research on the issue of unmanned driving, and the Ministry of Industry and Information Technology website has officially released the "Guidelines for the Construction of the National Standard System for the Internet of Vehicles Industry (Intelligent Connected Vehicles) (2017)" (solicited Opinion Draft) Notice of Opinions", a lot of opinions have been preliminarily established.

Whether it is to overcome technical difficulties or fasten safety ropes, the field of intelligent vehicles always requires a lot of investment of capital and talents. As Tian Feng said with emotion, the real economy represented by manufacturing industry is upgrading. The replacement process requires a lot of investment in research and development, machinery and equipment, and the introduction of intelligent robots. However, the row upon row of high-rise buildings takes up more and more tangible capital, and even pulls a large number of talents into the "battlefield" of cement and steel bars. On the other hand, this is undoubtedly a blow and a loss to a manufacturing industry like smart cars.

According to public information, the overall forecast shows that the total car sales will continue to grow , and the penetration rate of smart cars will become higher and higher. In any case, although there are many difficulties on the road to build a car, the day of its outbreak, sooner or later, will eventually come.

Photo / ZTE Smart car, singularity car