news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

the national development and reform commission predicts that by 2030, the penetration rate of smart cars in my country will reach 95%.

as wang chuanfu, chairman and president of byd group, put it, "the great transformation of the automobile industry, the first half is electrification, and the second half is intelligence." this trend is particularly evident at the 2024 chengdu auto show. all the participating automobile companies are working hard to show the world their latest achievements in the field of intelligence.

with the acceleration of the intelligent automobile industry, creating a complete set of intelligent network evaluation standards has become a prerequisite for promoting the rapid and healthy development of the entire industry.

(left: 58 auto editor-in-chief ji lei; right: zhang lu, deputy director of the technical management department of china automotive evaluation)

in this regard, 58 auto exclusively interviewed zhang lu, deputy director of the technical management department of china automotive evaluation and research institute, during the 2024 chengdu auto show, to discuss topics of concern to users, such as the future core competitiveness of smart cars, current technological breakthroughs, diversified testing scenarios and evaluation systems.

the core of intelligent driving technology lies in perception, decision-making and control

in the rapidly developing wave of science and technology, smart cars, as a revolutionary force in the automotive industry, are gradually changing people's perception and habits of transportation.

in the past, when consumers bought cars, they tended to focus on price, design, fuel consumption, power and other aspects. however, in addition to the factors just mentioned, current users also list intelligence as one of the important decision-making indicators when buying cars.

therefore, for oems, if they want to create a car with explosive product power, they need to pay attention to all aspects of the performance of smart connected cars, such as safety, comfort, experience, efficiency, etc. there are many points that can be deeply explored and optimized.

"take intelligent driving technology as an example. from the perspective of technical logic, its core lies in perception, decision-making and control. perception performance is the most core problem because it is closely related to the user experience. one problem here is that the user experience logic is whether i can use it in all scenarios. but in fact, current products are implemented step by step, and are still in the game period of performance and cost."

regarding the problems and controversies faced in the current development of intelligent driving technology, zhang lu gave three suggestions:

improve perception accuracy and reliability, especially in severe weather and complex environments. at the same time, focus on data fusion and optimize multi-sensor data fusion technology to ensure more accurate object detection and recognition.

improve the intelligence level of decision-making algorithms, improve the system's decision-making ability through more advanced algorithms, enable it to handle more complex and uncertain situations, and enhance the system's understanding of the surrounding environment, including traffic rules, road culture and driver intentions.

improve the robustness of the system to ensure that the system can diagnose and take redundant measures in a timely manner when a failure occurs to ensure safety.

“with the continuous development and cross-domain integration of intelligent connected vehicle technology, we believe that the development of v2x technology, artificial intelligence, and end-to-end technology will have a disruptive impact on the automotive industry in the next few years.”

the a and b sides of the smart car data “race”

it is said that data is the "oil" of the information age. nowadays, the massive amount of data generated by cars during use has become the most valuable asset of many car companies. therefore, a competition for collecting "car data" is quietly taking place among car companies.

according to statistics, tesla users drive more than 2 million miles in fsd mode every day. this massive amount of data can be used to help tesla train and improve the fully autonomous driving system.

large amounts of data collection and training make smart cars more and more "smart", but the other side of the coin is that it comes at the expense of users' personal privacy.

according to foreign media reports, an american hacker once stole data such as call records, home and work locations, navigation routes, and session records of allowed access to websites from tesla's in-vehicle computer system, which once again touched the nerves of users to protect their privacy and security.

in zhang lu's view, automotive data is driving the development of smart cars in an unprecedented way and has become the core driving force for innovation and progress. however, user privacy and security is the bottom line that car companies must adhere to and a red line that cannot be crossed.

based on this, china automotive evaluation and research institute added the world's first intelligent connected vehicle privacy protection evaluation system to the 2024 version of the china intelligent connected vehicle technical specification (c-icap) which was officially implemented on july 1. the system assesses test vehicles from two dimensions: network communication protection (anti-intrusion and tampering) and personal information protection (anti-voyeurism and theft).

by looking through the rules, we can see that the protection of personal information on the vehicle and mobile phone, such as in-vehicle video image information, in-vehicle audio information, vehicle location information, web browsing history, personal mobile phone number, license plate number, mobile phone control app, etc., are all included in the scope of vehicle safety assessment. it can be said that a "tight ring" has been put on the user information security issue of smart cars.

"the privacy protection system we have established is to tell users what their privacy is on the vehicle side, because sometimes users know what privacy is, but don't know what privacy they need to protect. similarly, car companies also need a set of such testing specifications to understand how to build an information security defense system. in fact, many car companies have invested a lot of energy in this kind of security protection, but users cannot see it. our set of regulations can reflect what level your system is at."

safety is the bottom line of all evaluation procedures

"in fact, our center has a main line in conducting intelligent network test, that is, safety is the first priority. for example, the principle of c-icap is safety first, and on the basis of safety, we also take into account the balance of comfort performance and related performance."

as we all know, the complexity of china's road traffic environment is much higher than that of other countries. therefore, special algorithm model training is required for some special road traffic scenarios.

to this end, china automotive evaluation corporation has built a large database of chinese driving scenarios, and has currently collected more than 3 million kilometers of real driving scenario data. by analyzing the frequency, complexity, degree of hazard and other characteristics of the test scenarios, they are included in the scope of each type of test scenario.

"after the evaluation, the performance of many models has been significantly improved, but there are still obvious deficiencies in the design of some corner scenes, such as ghost peeking, road construction scenes, etc. so in this regard, we actually have a lot of designs and considerations, and we also have a lot of data sources. after analyzing some features based on the database, we build these tests."

taking into account that some drivers have incomplete knowledge of vehicle functions and misuse or abuse intelligent assisted driving functions, which leads to traffic accidents, china automotive evaluation has also set up a special system description evaluation item in the evaluation procedures to examine whether companies explain the usage boundaries and inapplicable scenarios of their assisted driving functions when users purchase the car.

it is worth mentioning that in view of the differences in performance of intelligent vehicles in different regions and under different road conditions, china automotive evaluation corporation is considering establishing diversified testing scenarios and evaluation systems to more comprehensively reflect the actual application capabilities of vehicles.

"in the process of formulating the technical regulations for the c-icap assisted driving unit, we fully considered the functional definition and application scenarios of the intelligent system, which are divided into basic driving assistance and pilot driving assistance. in the future, we will also consider urban pilot. according to different intelligent systems, we will focus on the screening scenarios. for example, for the assisted driving system developed for highways, we consider more highway scenarios. for the assisted driving system developed in urban areas, we have introduced more scenarios of vulnerable traffic participants and intersections."

in the process of formulating the technical regulations of c-icap intelligent cockpit unit, the evaluation under different regions and road conditions was also considered. for example, in the voice interaction evaluation, china automotive evaluation set up voice interaction corpus for testing based on regional accents or dialect characteristics; at the same time, two noise interference scenarios were set up, urban roads and highways, to test the wake-up success rate and recognition success rate of the voice interaction system.

zhang lu said that in the process of formulating and revising the next version of the regulations, the impact of complex vehicle use environments caused by regional differences and road conditions on the use of smart cockpit functions will be further considered and incorporated into the evaluation system in order to more comprehensively reflect the actual application capabilities of the vehicle. "no matter which evaluation regulations are used, safety is our bottom line."

today, intelligence, as the second half of the automobile industry revolution, has sounded the "bugle" of the game. only by adhering to the bottom line of safety can car companies allow consumers to truly embrace the future of smart cars with peace of mind.