In recent years, the rapid evolution of intelligent driving technology has ushered in profound structural changes in the automotive industry. Autonomous driving has gradually evolved from the early assisted driving system (ADAS) to the new generation of production models that focus on "high-end intelligent driving" functions, and many car companies are competing to launch "L2++" and "quasi-L3".and so on to attract consumers. However, with the rapid advancement of technology, a series of problems that need to be solved urgently in the industry have also been exposed: the actual performance of intelligent driving functions does not match the publicity, the lack of standards, the lack of user education, and even the safety risks.
In this context, the "jizhi changce" led by the China Automotive Engineering Research Institute (hereinafter referred to as "CAERI") was developed and implemented, that's where the project comes in. The project is China's first standardized field long-term testing system focusing on high-end intelligent driving systems, aiming to reveal the actual performance of vehicle intelligent driving performance with real road environments, scientific evaluation methods and authoritative expert teams, so as to set a unified evaluation scale for the industry, help consumers make rational decisions, and promote the healthy development of technology.
Ⅰ Chaos in the intelligent driving market: excessive packaging and technological gap coexist
According to the 2023 J.D. Power China Smart Car Experience Study, more than 70% of users pay attention to the practicality and safety of autonomous driving functions in daily car scenarios, especially in complex urban roads, bad climates and unexpected road conditions. However, at present, many mainstream brands frequently use words such as "high-end intelligent driving" and "full-scene NOA" in their publicity, but fail to fulfill their promises in actual use.
For example, some models have problems such as misjudging traffic lights, identifying lane deviations, and frequently taking over in urban NOA (navigation-assisted driving) scenarios. What's more, due to the lack of authoritative standards and systematic evaluation systems in the industry, consumers can only rely on unilateral promotional information from manufacturers to make car purchase decisions without understanding the actual performance. This "information asymmetry" not only harms the user experience, but also limits the sustainable development of the industry.
Ⅱ "Extremely intelligent long test": build a scientific standard system to solve the evaluation dilemma
1. Multi-party collaboration to create authoritative evaluation specifications
Led by CAERI, the project took one year to formulate and complete the "Road Test and Evaluation Regulations for Advanced Intelligent Driving Assistance Systems" in conjunction with top domestic universities and technical institutions such as Tsinghua University and Tongji University. The regulations clarify the typical road scenarios, evaluation index system, quantitative scoring methods and test processes required for evaluation, covering the three core dimensions that users are most concerned about: scenario adaptability, regulatory compliance and behavioral anthropomorphism.
The evaluation team is composed of more than 30 technical experts, with interdisciplinary backgrounds such as intelligent driving algorithms, perception fusion, and vehicle dynamics, to ensure that the evaluation process is both rigorous and has industrial landing value.
2. The first test in Hainan: the actual combat test in a harsh environment
In 2024, the first round of "Jizhi changce" will be selected in Hainan for field testing. The region has unique climatic and geomorphological characteristics: high temperature, high humidity, high salinity and rainfall, which can effectively simulate the operation performance of intelligent driving systems under extreme working conditions.
This round of testing covers a total of 20 mainstream smart models in the market, covering independent and joint venture brands, and the test content includes more than 20 typical scenarios such as highway NOA, urban road recognition, S-curve control in mountainous areas, automatic parking, multi-vehicle interchange, and rainy conditions. Among them, the tropical rainforest road scene is very challenging to interfere with camera recognition due to strong light and shadow changes. For example, sunlight dappled through forest gaps, and some systems can misjudge the edge of the lane or even trigger misoperation, which is rigorously recorded in the evaluation.
Figure: The first standardized long-term test "Jizhi changce" opens the high-level intelligent driving long-term test research and test
3. Accurate data collection and quantitative evaluation
In order to ensure a high degree of credibility in the test results, CAERI is equipped with a self-developed on-board data acquisition platform with centimeter-level positioning accuracy and millisecond-level data recording capabilities. During the test, the system will continuously record the response data of each vehicle in various scenarios, including core indicators such as acceleration and deceleration delay, braking strength, direction correction times, takeover frequency, and false positive rate, and combine the video stream with map matching to form a complete behavior data package.
The data is scored by a unified algorithm and scored according to functional modules and scene classifications, so as to achieve highly objective scoring results. This can not only reveal the strengths and weaknesses of different models in subdivided scenarios, but also provide a clear direction for technology iteration.
Ⅲ From enterprises to users: the triple value of "jizhi changce".
1. Force enterprises to iterate on technology and purify the market environment
In the past, some manufacturers have seized the high ground of public opinion with the help of vague publicity terms such as "L2+" and "all-scenario intelligent driving", and put them into the market before the function is mature, overdrawing the trust of users. Based on a standardized evaluation system and real test scenarios, the "jizhi changce" will prompt manufacturers to return to their technical origins and improve the standardization of product development and verification.
After the evaluation results are made public, those "pseudo-high-end intelligent driving" models that rely only on packaging and marketing will lose their appeal to consumers; The technical strength brand with real performance advantages will gain a good reputation and market dividends. This mechanism has a significant role in promoting the overall technical level of the industry.
2. Provide consumers with decision-making reference and enhance security
One of the highlights of the "jizhi changcet" is evaluation process and results are open to the public. The scores and performance of each model tested in various scenarios will be transparently released. For ordinary consumers, this is equivalent to obtaining a set of car purchase "manuals" verified by authoritative institutions.
For example, for users living in first-tier cities, priority can be given to models with good acceleration and deceleration control and smoothness in congested road conditions; For users who often travel by car, they can pay attention to high-speed NOA continuous lane changes and models with strong ability to identify construction areas. This kind of "scenario-matched car selection" thinking is expected to replace the traditional "pile parameters, watch ads" car purchase logic, and significantly improve travel safety.
3. Enhance the international influence of China's intelligent driving
In addition to serving the domestic market, CAERI also plans to expand the "Jizhi changce" to countries along the "Belt and Road" and BRICS countries. At present, automotive regulators in many countries in Southeast Asia have expressed interest in the standard system. In the future, the evaluation system is expected to become an important starting point for China's export of intelligent driving technology standards, and help Chinese enterprises improve their credibility and competitiveness in the global market.
At the same time, the "jizhi changce " can also be used as one of the ways for China to formulate the right to speak on international standardization, and to establish the right to speak on technology and market rules in the context of immature global intelligent driving regulations.
Ⅳ Conclusion: A key step towards standardized intelligent driving
The "jizhi changce" promoted by CAERI is not only a technical evaluation project, but also a far-reaching innovation in the industry governance mechanism. It fundamentally fills the gap of the lack of standards in the field of intelligent driving, and injects a foundation of trust into China's intelligent driving industry with a scientific, rigorous and reproducible test system.
In the future, with the participation of more batches of models, the evaluation results will gradually form a big data platform to help the industry establish behavior prediction models and safety risk early warning systems. The long-term significance of the "jizhi changce" lies not only in the evaluation itself, but also in the fact that it will help the industry enter a new era dominated by standards and technology is king, and lead China's intelligent driving to a healthier, rational and international development path.
