Autonomous Driving 2022: Delivery capabilities will be a litmus test

Image source @ Visual China

Text | Tabor.com, author | Yang Wenfei, Editor | Shikino

Autonomous driving is difficult to develop. Despite the rush of technology and capital, the enthusiasm of the market has not diminished, and the commercialization of autonomous driving is still in a dilemma.

In recent years, intelligent vehicles have continued to exert efforts in assisted driving and vehicle-road coordination. At present, intelligent cars equipped with driver assistance systems have gradually become the "standard" of mid-to-high-end cars, and the penetration rate of the new car market has exceeded one-third.

As the function of implementing L2 autonomous driving became more common, this high-tech technology gradually lost its original brilliance. The public is beginning to turn its attention to more mysterious and sophisticated high-level autonomous driving.

Unlike Level 0 and Level 1, every upgrade after L2 means a leapfrog development of autonomous driving. Autonomous driving above L3 is still a deep sea waiting for further exploration in the industry.

Regarding the conjecture of the commercialization of high-level autonomous driving, we chatted with several industry insiders.

The "Three Mountains" of Autonomous Driving Commercialization

Autonomous driving is on the right path may be the consensus of the industry, but like layers of wrapped cocoons, when to break through the shackles to achieve advanced automatic driving, no one dares to guarantee.

"At present, passenger cars with consumers as the main purchase generally only have L1 or L2 level assisted driving functions." Mr. Gao of FAW R&D Institute said, "Vehicles with L4 and above driving functions are mainly some Robotaxi and driverless minibuses landing in some closed or semi-enclosed parks. This kind of traffic scene is relatively simple and the speed is relatively low. ”

When asked about the reasons why this phenomenon is difficult to break through, most of the respondents mentioned factors that are first of all unclear about the formulation of regulations.

Mr. Xi Ning, General Manager of HERE Greater China, said: "The regulations on autonomous driving in various countries are relatively lagging behind the development of the industry. L3 involves human-computer interaction, what scenario to start automatic driving, what scenario to take over by the driver; L4, L5 level of automatic driving rights and responsibilities division, are difficult problems. ”

"Compared with foreign countries, our country's advantages lie in strong infrastructure construction capabilities, centralized planning, and a high degree of digitization, but it is more stringent in terms of regulations, including encryption offsets, compliance and map review requirements, while the population density is extremely large, and the road conditions are more complex."

Mr. Li Yang, R&D Director of NavInfo's new ADS team, holds his own views on this issue. He believes that the development of autonomous driving is a slow and long-term process, and regulations and policies and the industry promote each other. If a product wants to survive in society steadily, it and management must have an organic symbiotic relationship.

"When the strongest people in the practitioners have not really made it, can serve tens of millions of homes, and let everyone know that the system can be foolproof, it is difficult for regulations to give clear standards." It's hard to ask policymakers to know more about autonomous driving than technologists, and it's a mutually reinforcing process. ”

At a strategic level, the improvement of safety management of autonomous vehicles is indeed slow.

In April 2021, the Ministry of Public Security solicited public opinions on the Road Traffic Safety Law (Revised Draft), which has added new management requirements for autonomous driving, including requirements for road testing of vehicles with autonomous driving functions, and the determination of liability for illegal acts and accidents of automobiles with autonomous driving and human direct operation mode.

Secondly, the difficulty of breaking through technical barriers and the slow evolution of knowledge systems are factors that limit the development of automation.

"Autonomous driving is a systems engineering, and to achieve full-scenario use, the requirements for perception, software and hardware are very high." Mr. Xi Ning believes that high-precision maps and real-time dynamic services as extended sensors are very important for improving the perception and safety redundancy of autonomous vehicles.

"Only with the development of sound regulations, more mature recognition and perception technology, and the integration of software and hardware of control systems can autonomous vehicles surpass the capabilities of mature human drivers." These are all core elements of the development of autonomous driving." Mr. Xi Ning said.

Technology is difficult to break through, and costs are correspondingly difficult to reduce. Commercialization must be related to cost and original value.

In recent years, the application of cameras, millimeter wave radar, and ultrasonic radar technology in automobiles has become more and more mature, and the price has been declining, and the industry is optimistic about the significant reduction in the cost of autonomous driving.

According to statistics, at present, the unit price of the vehicle camera continues to decline, currently about 150 yuan, it is expected that the future price reduction space is limited; the market unit price of millimeter wave radar is about 600 yuan, and there is still a certain room for price reduction in the future; in recent years, solid-state lidar and other products have begun to replace traditional mechanical radar, and the process cost has dropped significantly, coupled with the future development of automatic driving technology to drive the supply of supply, there will be a large space for price reduction.

"The higher the order of maturity of the market, the lower the cost. If the country produces tens of millions of autonomous vehicles every year, the cost will be reduced again. But it's a lot cheaper now than it was five years ago," says Mr. Li Yang.

At present, high-level autonomous driving is trapped in the scene.

At present, the application scenarios of automatic driving are mainly based on high-speed trunk lines, terminal distribution, mining areas, ports, etc., as well as sanitation and cleaning. These scenarios are closed, and it is easy to realize autonomous driving technology. But even as a just need, the limitations still exist: the market demand is too small.

"Passengers in first-tier cities use commercial software to hit Robotaxi, most of them are to taste the fresh", an industry insider said, the current sense of driving experience brought by the car experience, and the mature human driver is still far behind.

In addition, he added: "Under the premise that the driver's driving technology is mature enough, some scenes can be driven by means of improvisation; but in order to ensure safety, most of the strategies of autonomous vehicles are more conservative, and they can only wait for the car in front of them - the car in front does not move, and the driverless car will wait all the time." ”

As a new thing, the market acceptance and demand for autonomous driving are not high. Rather than waiting for it to land, should we think about it: Are we really ready for it?

In the dark phase, accidents are difficult to avoid

Recently, Tesla said that the company's Autopilot automatic assisted driving has made driving safety 8.9 times the US average.

In August 2021, a domestic car owner driving the NIO ES8 had a traffic accident on the highway and unfortunately passed away, using the automatic driving assist function when the accident occurred. In September, a owner of Xiaopeng Automobile opened the NGP on the highway and collided with a cargo board in front of him, and the owner was hospitalized for concussion afterwards. The Xiaopeng owner said: "I thought the NGP could automatically identify the scooter and avoid it, but when I found out that it couldn't work, the brakes were too late." ”

Is the autonomous driving function really safe?

"First of all, when studying the instruction manual of the car, we will find that if the function is to be turned on, the traffic environment of the car must be good. Most of the traffic accidents that occur now are in a bad environment, or the car cannot recognize obstacles at all. In this situation, accidents are difficult to avoid. Consumer usage scenarios are complex, and existing technologies do not fully cover all scenarios. Mr. Gao believes that at present, automatic driving is still in the stage of moving forward in the dark, and it is difficult for us to avoid accidents.

"Autonomous driving is so new that people are too optimistic about its early understanding." Mr. Li Yang believes that when the theory is continuously built, problems will be found when it is verified, and it will experience a trough when adjusting, and the development of automatic driving will have a dormant process.

"From full of twists and turns to gradual clarity, this is the natural process of the development of things," he said. ”

For frequent accidents, Elon Musk said in an interview with Time: "When we started to pursue self-driving technology, someone said to me: Even if you save 90% of your life, the 10% of people who fail to save will sue you. ”

"So I think you don't necessarily get rewarded for saving lives, but you're going to be blamed for not saving lives."

Every year is the "first year" of autonomous driving

At a time when there are more and more vehicles and roads are becoming more and more crowded, self-driving cars are undoubtedly the trend of the future. Some analysts believe that 2022 is a watershed year for the commercialization of autonomous driving.

"As a rapidly developing new technology, autonomous driving has made breakthroughs every year, and every year can be said to be the first year. 2021 is also a year with many highlights, Mercedes-Benz's L3-level autonomous driving system was approved by the German Federal Motor Transport Administration on the road, becoming the world's first mass-produced L3-level autonomous driving platform on the road. Ms. Chen Yan, Chief Architect of HERE, thinks.

In her view, the research and development of automatic driving is a process of accumulation and thin development: "China's automatic driving market potential is huge, every car company and automatic driving players also have their own solutions and roadmaps, in my opinion, every year is a new starting point for automatic driving technology, technology continues to innovate, digital and intelligent applications develop rapidly, China's car companies and autonomous driving players will create more milestones in the entire industry." ”

Mr. Li Yang believes that 2022 is not so much the first year of the commercialization of automatic driving, but rather the first year of automatic driving delivery.

"If you look at the initial signals of a wave, 2021 has given ample signals to prove which is the origin and which is the watershed. In 2022, it is more to wait and see whether mass production can be delivered. ”

"In the past, car companies were willing to release opportunities for domestic suppliers, or rather, car manufacturers were willing to make great strides into L2+ and approach the L3 field, so everyone got orders in 2021." By the end of 2022, if the supplier is unable to deliver on time, the project may be canceled or the supplier replaced. ”

In the short term, the commercialization of advanced autonomous driving may still be trapped. When it comes to the ultimate autonomous driving, most practitioners are not optimistic. Even, some people think that it will not necessarily come.

"Because the cycle is too long, we can only talk about autonomous driving in specific business contexts. At present, the commercial motivation for autonomous driving is sufficient, and if there is no possibility, no one has the motivation to do it. I think it will take at least 5 years for autonomous driving to usher in a positive business situation based on cost. It is impossible to judge in the near future. ”

Li Yang believes that he prefers to see the value of autonomous driving itself than to show radical technology.

"To promote the commercialization of autonomous driving, we should focus on the product value and social value brought by automatic driving itself, and do a solid job in products, rather than simply pursuing the cool of automatic driving." People who do automatic driving probably have a bigger future in their hearts: by liberating drivers to reduce traffic accidents, to save people, improve efficiency, realize traffic intelligence, and truly make social value. ”