Bin Liu, Policy Research Center, Institute of Automotive Technology Information, China Automotive Industry Corporation
In order to better promote global cross-border collaboration, clarify the role of travel agents, automobile companies, cities and other subjects in the future travel ecology, and discuss major changes that will occur in the future travel methods, travel technology, transportation structure, and transportation strategy As well as the opportunities and challenges that the entire industry will face, China’s Hundred People’s Association of Electric Vehicles organizes the 2019 Global Future Travel Conference to promote travel ecological transformation, transformational innovation, and international collaboration to welcome the travel revolution. The following is a record of Liu Bin ’s speech at the Policy Research Center of the Automobile Technology Information Institute of the China Automotive Industry Center:
Before a hundred people would give me this topic, I also conducted in-depth research. Today, Zhang Chu is an expert in this field. We previously worked with the Development and Reform Commission on smart car strategy because the center is mainly the automobile standard to assist the Ministry of Industry and Information Technology. We are different in this process Share the technical route, some discussions, and the importance of high-precision maps.
My sharing is divided into four aspects: First, the current status of the development of high-precision map industry. Second, there are some problems facing the industry. Third, policy trends. Fourth, we think about this. We used to be researching automobiles and later were new energy vehicles. Intelligent network integration has just been involved, so it may not be correct. It is just some academic research for everyone to share.
First of all, although it is said that future autonomous driving will have different technical routes, high-precision maps are necessary for different technical routes of autonomous driving. We SAE has divided five layers into autonomous driving. The front L1 and L2 may not be needed, because mainly human driving, the map reference is sufficient. At the level of L3, it is necessary to provide real-time dynamic traffic information, and it is necessary to provide the vehicle’s environmental awareness while navigating the map and the sensor mode. As for the sensors, the previous experts also said that the requirements may be relatively high. The combination of high-precision maps and sensors can reduce the cost of sensors and facilitate subsequent mass production.
What is the specific situation of the industry? We analyzed it, and foreign companies probably entered the map merchants directly through acquisition and cooperation, including BBA, GM, including Japan, and various inspections directly entered. Due to safety requirements in China, major vehicles in Shanghai cannot enter. For this reason, it is mainly for Class A surveying and mapping qualified enterprises to make high-precision maps. Domestic vehicle companies generally make arrangements through cooperation or acquisition. At present, there are mainly three companies with Class A qualifications. We have sorted out about 20 Internet technology companies are increasing their investment.
At the same time, the competent authority has also changed due to the adjustment of government agencies. It is now mainly affiliated with the Ministry of Natural Resources, mainly the Department of Land Surveying and Mapping, the Department of Geographic Information Management, and other departments and departments. Policies, audit qualifications, supervision, and more.
At present, because I am doing policy research, I mainly look at the possibility that a high-precision map has formed a three-dimensional management policy system based on laws and regulations, and combined department regulations and standards. What has been blackened are some regulations that we think may be more important. Everyone can pay moderate attention.
With the development of dynamic driving, there may still be some problems with high-precision maps in the development of the industry. We have sorted out the process of making high-precision maps. There are about six aspects. Many of you present may be experts, not necessarily correct. First, the data collection mode includes centralized and crowdsourcing. Second, sensitive information may need to be eliminated. Third, generating map products is mainly performed by map dealers. Fourth, the technical processing of the spatial position of the map. Fifth, government departments conduct audits. Sixth, use maps.
Problems that may be faced during these six steps:
First, the qualification of high-precision maps. Anyway, as we all know in the automotive industry, the qualification of the Development and Reform Commission is the first threshold and it is relatively strict. For national security considerations, the qualification review is strict. At present, when producing maps for autopilot maps, car navigation companies are mainly engaged in the independent surveying and mapping activities by electronic map production units. At the same time, foreign-funded enterprises are not allowed to engage in the production of navigation maps.
Second, there is a regulatory gap in the current crowdsourcing model. Because autonomous driving maps require high timeliness, relying on professional vehicle collection time is too late and cannot be updated in a timely manner. Now some companies are conducting crowdsourcing collection, including collecting location information, vehicle information, and sensor information. However, for crowdsourcing, we have not established such a regulatory system, nor explicitly stipulated which data can be collected during crowdsourcing, and there is no stipulation on how to ensure the security of data security backhaul, and there are no clear requirements for the qualifications of corporate entities.
Third, China forbids marking sensitive information on maps. Including the road’s maximum breakage, maximum (small) curvature radius, this is needed for autonomous driving height. For example, we usually use navigation, maps, and roads. If you are on an elevated road, you cannot tell whether you are on an elevated road. It may be a two-dimensional map, and it may be difficult for autonomous driving to rely on it.
Fourth, the deflection encryption before the map is published will affect the accuracy of the map. We have related technical requirements. The security scheme of the map encrypts the data of high-precision maps, and the satellite positioning signals also need to be deflected, which brings challenges to high-precision positioning.
Fifth, the current navigation electronic map review mechanism brings challenges to the real-time update of high-precision maps. This is not much to say, it is mainly a real-time issue, and the review process is now relatively long.
After researching the policy trend, we may think that we must not only ensure security but also encourage innovation. Policy formulation may adhere to three principles: first, to maintain the bottom line of national security, and to exchange geographic information data related to national security. , Sensitive geographic information data must keep to the bottom line of security. Second, in consideration of the development of innovation, we should give some support to innovation, including the fact that we now have some relevant policy support in Zhejiang and Deqing. Third, the principle of phased, step-by-step, to relax restrictions that hinder industrial development. But this process may still be a bit long.
There may be four trends in policy trends: First, the approval of qualifications, in order to accelerate the development of the industry. Second, whether the encryption plugin upgrade can be speeded up. Third, the restrictions on the expression of information, whether it can be relaxed, implicit expression, and individual important releases. Fourth, can the map review process be accelerated, especially the optimized high-resolution map review process?
Finally, there is a brief consideration: including release, such as whether it is possible to release the restrictions on expressing the slope and elevation of roads in the map on sensitive information processing, and if there are some technical means to deal with if it cannot be released.
It is just a matter of conjecture to consider whether the map accuracy limit can be relaxed. 1. Thoroughly release data confidentiality requirements. 2. Properly relax the requirements for encryption accuracy, plus use some technical means to improve the detection accuracy of the vehicle, for example, using a high-precision map plus strong perception. 3, do not let go of accuracy requirements, mainly rely on strong perception, equivalent to a vehicle with a very high degree of braking, automatically identify the road to travel. 4. In combination with the characteristics of high-precision maps, establish a corresponding map review mechanism to clarify what and how to review, while also considering documentary requirements.