Public security organs have issued a total of 16,000 test plates for autonomous vehicles and opened 32,000 kilometers of public test roads. How far is China's autonomous vehicle industry from true commercialization?

Intelligent and connected vehicles, commonly known as driverless or autonomous vehicles, are currently a major high ground in the global technological revolution. Recently, there have been numerous positive developments in the field of autonomous driving, with both technological breakthroughs and initial signs of commercial applications, reigniting widespread discussion in the industry.

Amid the continued heating up of autonomous driving technology and heated discussions in the industry, the State Council Information Office held a series of press conferences on the theme of "Promoting High-Quality Development" on August 27. According to reports from Auto Insight, China's intelligent and connected vehicle industry is advancing steadily through continuous technological accumulation and testing development. Notably, public security organs have also demonstrated a proactive stance in promoting the implementation of autonomous driving technology. As of August 2024, China has issued a total of 16,000 test plates for autonomous vehicles and opened 32,000 kilometers of public test roads.

These measures have provided strong support for the validation and iterative updating of autonomous driving technology. However, the process of commercializing China's autonomous driving technology has not been smooth sailing. While some enterprises have taken brave steps forward in commercial applications and mode innovation, the overall business model still needs further exploration and optimization. Additionally, policy support and regulatory construction are also key factors restricting the development of the industry. During the 2024 China Automotive Forum, many industry insiders called for the need for not only efforts from enterprises across the industrial chain but also support from infrastructure, laws, and regulations to achieve full-scale commercialization of autonomous driving technology.

Collaborative Industrial Chain, Capturing the "High Ground" of the Industry

Good news is that the national level is actively promoting the improvement of relevant laws and regulations to address the new challenges posed by autonomous driving technology. According to Wang Qiang, Director of the Traffic Management Bureau of the Ministry of Public Security, the Ministry is currently working on revising the Road Traffic Safety Law, which has been included in the 2024 legislative plan of the State Council and the first-category project of the 14th National People's Congress Standing Committee's legislative plan.

On the other hand, public security organs are actively cooperating with industry regulators to promote road testing and are currently establishing a comprehensive management system for autonomous vehicles covering road testing, demonstration applications, market access, and road travel.

In terms of road testing and demonstration applications, the Ministry of Public Security, the Ministry of Industry and Information Technology, and the Ministry of Transport jointly issued a document in July 2021, specifying the entities, drivers, and vehicles involved in road testing and demonstration applications of autonomous vehicles, as well as requirements for road testing and demonstration application management, traffic violations, and accident handling.

Regarding market access pilots and road travel, the Ministry of Public Security collaborated with the Ministry of Industry and Information Technology in November 2023 to establish regulations for pilot market access for autonomous vehicle products that have undergone technical testing and are ready for mass production. Since 2024, cities such as Beijing, Shanghai, Shenzhen, Guangzhou, and Wuhan have been promoting research and development and applications of driverless technology through policy guidance, capital investment, and infrastructure construction, aiming to capture the "high ground" of advanced intelligent driving in the industry.

Among them, Beijing's High-Level Autonomous Driving Demonstration Zone issued notices for high-speed road manned demonstration applications to companies such as Baidu and Pony.ai in February 2024. In July, Shanghai issued the first batch of demonstration application licenses for fully autonomous intelligent and connected vehicles, allowing their use on certain sections of roads in Pudong, Shanghai, without drivers. Additionally, Guangzhou recently opened the first batch of high-speed intelligent and connected vehicle test routes, providing a window for autonomous vehicle testing and validation.

"End-to-End" Revitalizing Autonomous Driving

In terms of technological breakthroughs, the emergence of "end-to-end" large model applications has undoubtedly painted a revolutionary blueprint for the autonomous driving industry. Industry insiders generally believe that end-to-end large models will significantly shorten the transition time from automated driving assistance to fully autonomous driving. Xiaopeng Motors' Chairman, He Xiaopeng, predicts that within the next 36 months, the application of end-to-end large models in autonomous driving will enable everyone to drive like an experienced driver in every city.

Among automakers, Tesla was the first to apply the "end-to-end" large model to mass-produced vehicles. This innovative measure is embodied in its FSD V12 version and has received widespread industry acclaim. Entering 2024, companies such as Xiaopeng, NIO, LIXIANG, ZERO, ARCFOX, Huawei, and SenseTime have actively followed suit, introducing end-to-end autonomous driving solutions and models targeted at mass production.

These solutions not only demonstrate strong competitiveness at the technical level but also prove their effectiveness through excellent real-road performance. Specifically, compared to traditional "perception-decision-control" intelligent driving systems, end-to-end models integrate the original multi-model architecture consisting of perception, prediction, and planning into a single "integrated perception and decision-making" model. Without intermediate rules, this approach offers advantages in information transmission, reasoning calculations, and model iteration, enabling stronger general obstacle understanding, over-the-horizon navigation, road structure comprehension, and more human-like path planning capabilities, significantly enhancing the intelligence of individual vehicles.

However, despite its enormous potential, end-to-end autonomous driving still faces numerous challenges in achieving mass production and widespread adoption, such as constructing the required powerful computing power, acquiring high-quality and massive data for model training, and addressing unresolved issues like "opacity" and "insufficient interpretability." These challenges require concerted efforts from all parties in the industry.

It is worth noting that automotive industry organizations and enterprises have already begun making efforts to address the above issues. During the recent 2024 China Automotive Forum, the "Open-Source Simulation Scenario Set Generated from Multi-Source Data Integrating Vehicle, Road, and Cloud" was jointly released by the China Association of Automobile Manufacturers, Shanghai International Automobile City (Group) Co., Ltd., Shanghai Cheyun Data Technology Co., Ltd., and Zhonglian Technology Co., Ltd., providing strong support for the testing and validation of end-to-end autonomous vehicles. With the combined efforts of multiple parties, true autonomous driving is no longer far away.

Note: This article was originally published in the "Hot Topic Tracking" section of the October 2024 issue of Auto Insight magazine. Please stay tuned.

Image: From the internet

Article: Auto Insight

Typesetting: Auto Insight