How far is autonomous driving from the masses? |BEYOND Expo 2022

On September 21, the opening ceremony of the second BEYOND International Technology Innovation Expo (hereinafter referred to as “BEYOND Expo 2022”) was held in BEYOND Metaverse. This time, BEYOND Expo and ICBC (Macau) will build a consumer technology summit, and discuss the application of innovative home technology, personal technology, travel technology and other emerging sub-fields with the world’s top technology companies and representatives, and discuss the application of new technologies to empower industry transformation and innovation. New opportunities, promote the transformation of consumer technology business, and look forward to the development trend of consumer technology.

Under the unprecedented changes in a century, autonomous driving is causing systemic changes in the automotive and travel industries around the world. China is striving to build a world-leading new energy and intelligent connected vehicle industry highland, with unprecedented opportunities and challenges. This time, Shen Yushan, a reporter from the English version of Dynamic Technology, had a dialogue with Yu Qian, co-founder and CEO of Qingzhou Zhihang, and Dr. He Tao, founder and chairman of Kuwa Robot. Starting from the top-level design, it aimed at the development and commercialization of autonomous driving technology. issues are discussed in depth.

The bottleneck of autonomous driving

Five or six years ago, the entire automobile industry had not been able to be called a car with autonomous driving capabilities, and some models were only equipped with some functions for driving assistance, such as automatic cruise, lane keeping, lane departure warning and other functions. The whole is at the level of L1 and L2 in the automatic driving level division. Today, there are many products that claim to have “autonomous driving” functions launched by OEMs. L2 is standard, and L3 and L4 vehicles have also appeared.

Yu Qian pointed out that the commercialization of L4 will be implemented first in some vertical fields. The core of L4’s implementation is the security issue, followed by the cost issue. The commercialization of autonomous driving currently needs to solve some long-tail problems, which to a large extent must rely on massive data and be realized in a data-driven way. The acquisition of a large amount of data and the operation of ODD (Operation Design Domain) have put forward high requirements for the landing scene and business, so the data-driven closed-loop capability is very important.

“It will take 5-10 years for the widespread popularization of Robotaxi, which is related to the restrictions and scenarios of landing. Therefore, we propose a dual-engine strategy, using L4 technology to do advanced assisted driving to obtain more data and Scenario application. At the same time, the improvement in the capabilities of sensors and computing in recent years has also increased favorable factors for the widespread implementation of high-level autonomous driving technology. For example, the cost of lidar has increased from tens of thousands of dollars a few years ago to the current cost of hundreds of dollars Decline, the calculation has changed from a few TOPS (processor computing power units), dozens of TOPS to hundreds or even thousands of TOPS, and the concept of large computing power has emerged.”

He Tao said that in order for an emerging technology to truly achieve “revolutionary success” and bring transformative development to the industry, the rhythm is very important, that is, to continuously invest in various elements required by the technology in advance, so that the technology can continue to iteratively upgrade until we Breakthrough to break the minimum technical bottleneck it needs for this industry.

For the L4 problem in limited scenarios, some algorithm frameworks in the entire theoretical circle can be solved at present. There are two preconditions: First, some unpredictable long-tail problems should be avoided, and even some unpredictable long-tail problems should be avoided. Some difficult cases that were foreseen but could not be resolved in the short term.

“The data acquisition of Waymo, Baidu, etc. basically depends on the launch of test vehicles, and the essence of vehicle launch is purely capital-driven. Because the value of autonomous taxis as an industry may not exist for a long time, safety testers It is equivalent to a traditional driver, which exists in a long-term range, and redundant unmanned vehicle equipment, sensors and computing platforms must be placed on the car. Two systems of manned and unmanned coexist, but data can be obtained by this method. Therefore, its long-term development actually needs to be driven by burning money and depends on the financing ability of capital.”

He Tao believes that some companies do not use road test models or capital to obtain data, but try to use the value of autonomous driving products in vertical industries as much as possible, form production tools and production efficiency, and obtain certain profits from it. , and at the same time, the data can be collected by the way. “The way its data is obtained here is to output the product, but the essence of this depends on the maturity of the product, whether it can be accepted by this industry, and whether it can have a certain industry value in this industry. More consideration It is the company’s understanding of the industry. The second way is more suitable for some new start-up companies without strong capital support. By tearing open a vertical industry, find the autonomous driving technology required by the industry, and export this product to the vertical industry. Go inside the industry to get a steady stream of data.”

policy boost,Commercialization of autonomous driving accelerates

China is stepping up its pace to open the door to commercialization of self-driving vehicles: in July, Beijing Yizhuang self-driving taxis ushered in an unmanned commercial pilot, moving the safety officer from the main driver to the co-driver. Then Chongqing and Wuhan recently announced the direct cancellation of the pilot program of fully unmanned safety officers, and then to the introduction of the latest guide “Guidelines for Self-Driving Vehicle Transportation Safety Services (Trial)” – this is the first time in China that self-driving vehicles can also be used for transportation operations. , the scenarios that are clearly encouraged include closed express buses, taxis, trunk freight, port terminals, etc. “The policy has taken a very big step forward”, and autonomous driving is expected to really start commercial operation.

The release of this guide is of epoch-making significance in the autonomous driving industry, and it also makes autonomous driving practitioners very excited. Because this policy gives a very important step in the commercialization of autonomous driving – operation. This means that the operating conditions that were only opened in some local policies before can also be promoted in other cities, instead of just staying in the “testing” stage, but actually moving towards the stage of application. In essence, it promotes the rapid landing of commercialization.

“There is still a big gap between autonomous driving technology and business. From the perspective of Qingzhou Zhihang, we have always attached great importance to the balance between technology and business, and bridge the gap between technology and commercialization. Through a business closed loop, to promote a larger business closed loop and step by step to promote wider commercialization, this is a concept we have always adhered to.” Yu Qian, co-founder and CEO of Qingzhou Zhihang, believes that “the future of Qingzhou Zhihang is very important. Speaking of, there are two general directions for the commercialization of autonomous driving – the future space of L4-level Robotaxi will be very large, but it is very difficult, and it will take 5 or even 10 years to popularize, so the space is large, and many landing cycles will be very long. It is also very challenging; the application scenario of Robobus, which is oriented to the open road, is not fundamentally different from Robotaxi, but compared with Robotaxi, its ODD is smaller and the speed is relatively slower, so it can achieve a closed-loop business faster. So Our Robobus product can largely solve the pain points of short-distance multi-person travel caused by China’s large population base and high population density, which is very suitable for China’s national conditions.”

He Tao, founder and chairman of Kuwa Robot, said that the core demands of local governments are from the perspective of industrial development, combined with the characteristics of the city, to open up different autonomous driving test scenarios, which can not only introduce new industries, but also make Its original car-making or equipment-manufacturing industries can obtain secondary development. At present, policies related to autonomous driving in various parts of China have different priorities. Some places focus on industrial development, some places focus on the integrity of legislation, and some places focus on the integrity of supervision. Therefore, it varies from place to place, but gradually In the future, each place will learn from each other, and gradually it will expand from regulation to legislation to policies, complement each other, and become compliant.

“With regard to the dependence of industry innovation on autonomous driving technology, there is no doubt about sanitation and urban distribution. If an industry introduces autonomous driving technology, it will have a great impact on its traditional industry model, that is, autonomous driving technology has become a kind of Core and supporting technologies, we will attach great importance to the early commercial implementation of this industry, so we will pay special attention at the beginning, such as urban distribution, municipal sanitation and minibuses, because we think that from the industry innovation In terms of the dependence of this kind of automatic driving technology, once they introduce automatic driving technology, it will greatly change the production efficiency achieved by the original production tools of these two industries. Not only can replace human drivers to reduce costs, but also make bicycles more efficient. The operational efficiency and the resource scheduling efficiency of multiple vehicles have been greatly improved.” He Tao said, and he also admitted that Kuwa Robot focuses on sanitation, urban distribution and minibuses, and focuses on three verticals in urban dynamic scenarios. Subdivide the scenarios and actively promote the implementation of commercialization.

Top-level design to be optimized

The intelligent networked vehicle industry is developing rapidly. The existing guidance application process is cumbersome and the cycle is long, and the road test license plate cannot be entered into the system of the public security and traffic police, which has been difficult to match with the rapidly developing industry. Yu Qian believes that the relevant departments may be able to clarify the product catalogue on the road, and the license plate should be released first (the responsible subject is clear), and the information will be supplemented later. This will buy time for the formulation of the product catalog system, the approval of catalog access applications, and the application of license plates and driving licenses.

Second, clarify the standards for determining liability for traffic accidents. In the era of human-machine co-driving, the standards for determining liability for traffic accidents involving automatic driving systems. The driverless accident rule is very important. Once an accident occurs in the autonomous vehicle operation service, it will definitely involve the division of criminal liability and civil liability. Unlike traditional auto transportation services, the administrative department needs the rights of two new actors (the competent user and the authorized autonomous driving entity). Only by defining the responsibilities and clarifying the legal relationship, can the regulations be followed.

Third, expand the scope of commercialization charges. If the intelligent network connection industry is further mature, it will be unavoidable to explore in terms of commercialization and scale, and it will inevitably involve operating charges. However, it is difficult for autonomous driving companies to apply for operating qualifications; can you refer to Beijing, Shenzhen and other places , by applying for national or provincial authorization or legislation in the first area, to grant “road transport business license-level road transport certificates” to autonomous driving companies involved in commercial exploration, so as to avoid difficulties in implementing the original relevant norms. Pass.

Finally, he hopes to expand the application scope of autonomous driving technology/expand the scope of autonomous driving test roads, and promote the application of autonomous driving technology in urban public transportation and road passenger transportation. “The development of technology needs to be verified in a wider range of scenarios. By expanding the scope of the test road, it can help us to promote the technology to be verified more widely, and then support the exploration and development of autonomous driving public transportation demonstration applications in typical areas. , and based on the technological evolution and demonstration progress, on the premise of doing a good job in risk assessment and emergency plans, it will be extended to other passenger transport scenarios on public roads as appropriate. This is also what we very much hope to see.”

In response to this topic, He Tao believes that the current threshold regulations for the market access system should be set loosely, allowing more people to participate, first enter and then supervise, and then go through the process of survival of the fittest. Each company will bring its own innovations and its own different industry perspectives, all of which will help make the entire market more mature.

Secondly, he said that the opening of the scene is very important, but in the process of opening, the main body of the opening is actually the local government. “Autonomous driving often reshapes the entire industry chain, and it will definitely reshape the original interest pattern. If we want to break the original interest system pattern, we need local governments to stand on a more macroscopic dimension and a higher pattern, and at least create a fair environment.”

The future of autonomous driving

2025 may be a key node for autonomous driving. At present, we can see the landing of some leading autonomous driving companies such as Cruise and Waymo in the North American Bay Area. However, Yu Qian believes that the Chinese market and the US market will be very different, because the labor cost in the United States is higher, the road conditions are relatively better, and the road complexity is less than in China. Therefore, he believes that the landing of robotaxi in the United States will be faster.

“Of course, I also very much hope to see that the landing of robotaxi in the United States will play a better role in the development of the entire industry, and at the same time, it will also promote the large-scale landing of autonomous driving in China. I think this is a very good one. Prospects. In the field of autonomous driving, everyone may be pessimistic that some recent developments may not meet expectations, but in the long run I am still very optimistic about the large-scale implementation of high-level autonomous driving.”

He Tao pointed out that the era of autonomous driving relying on capital to obtain data is over. In the post-epidemic era, the most typical feature is that capital becomes more valuable. Investors will invest cautiously according to whether autonomous driving can really bring large-scale industrial effects and successful business models.

“From this point of view, it is not conducive to the development of autonomous driving. As for the current autonomous driving, especially the L4 autonomous driving, it has not yet reached a clear future for capital and a short-term effect. In this process , we will find that in fact, in the future, the autonomous driving enterprises, different landing enterprises may be differentiated, one is the robotaxi, from the general trend, the L4 level robotaxi may be more difficult to finance and the cost will be higher and higher. Second, for autonomous driving players developed for vertical industries, it may be more concerned with the depth of the coupling between technology and vertical industries, and whether a scale effect can be formed in the short term, which requires trade-offs, and we are constantly adjusting ourselves. rhythm and adaptability to the industry, that is to say, in which sub-sector is the current autonomous driving technology mainly cultivated, it is most likely to have a large-scale effect in the short term and form certain future expectations, so we will block all the Problems that are unpredictable and cannot be solved by L4 now. At the same time, we will further integrate existing technologies and implement them to form short-term production efficiency. For example, we value urban distribution and sanitation, because in the short term we will provide B-end customers and The industry forms an economic account, and estimates what kind of increment we will have in this market every year.” He Tao believes that in this process, the financing progress of the entire industry may need to rely on several companies to boost confidence in the capital market.

Leave a Comment