Tesla’s out-of-control incidents are frequent, but is it because of the crash caused by this system？

Every reporter | Li Xing every editor | Sun Lei Cheng Peng　　　　

A few days ago, Tesla’s out-of-control incidents occurred frequently, which caused the industry’s concern about the safety of autonomous driving data.

Take the owner of the accident in Zhengzhou, Henan Province as an example. At the end of April, under the order of Zhengzhou Zhengdong New District Market Supervision Bureau, Tesla provided the data of half an hour before the accident to the owner Zhang Jing (a pseudonym) and announced the data of one minute before the accident to the outside world.However, due to the incompleteness of the data it provides, Tesla has been questioned.

Some analysts believe that the complete data can analyze and judge whether there is a "brake failure" in the accident vehicle and whether there is a BUG in the Tesla system. "Before seeing the complete data, whether the vehicle caused an accident due to the failure of the’ brake’ can’t be concluded for the time being." A number of independent brand car companies related technical personnel said in an interview with national business daily.

In this context, Tesla also plans to launch a data query platform during the year. "The domestic vehicle owner data inquiry platform is under construction for all vehicle owners to inquire about the data of vehicle-machine interaction, and it is expected to go online this year." On May 11th, the relevant person in charge of Tesla China told the reporter.

Tesla "step on the pit" Linux time-sharing system?

"Until now, Tesla has not provided us with complete data." Zhang Wei’s husband told reporters that TeslaThe two data provided before are missing important parameters, such as brake pedal displacement, accelerator pedal displacement, motor torque and many other key data.

For the data released by Tesla, an executive of a self-owned brand car company who asked not to be named said that this data is incomplete. From the existing data, it is impossible to see what caused the accident.

Some people think that Tesla products have these problems because they use time-sharing operating system as the core system of vehicles.Time-sharing operating system will be unable to complete some tasks, and the tasks that cannot be completed will cause the system to crash.

Image source: photo by Zheng Derui, national business daily (data map)

"Tesla’s entire operating system is based on Linux, and it is equipped with a non-real-time (that is, time-sharing) operating system. From now on, Tesla has made some cuts to the Linux system, making the system response speed fast enough. "An executive of an automotive software technology company who asked not to be named told reporters,Linux is currently the best system to support the autonomous driving ecosystem.

According to Li Yuhao (a pseudonym), a senior computer programmer, neither real-time nor time-sharing operating systems can guarantee 100% security, which is mainly related to the reliability of the system.

"Time-sharing operating system is a processing task in which a time period is divided into time slices, giving priority to efficiency. If there are many tasks, real-time data monitoring may not be guaranteed. The real-time operating system actually has such a problem. "Li Yuhao told reporters.

For the time-sharing operating system used by Tesla, will there be a situation where the data cannot be monitored? Li Yuhao said: "If there is a BUG in the system, there may be cases where data cannot be recorded."

The above-mentioned senior executives of the automotive software technology company, who did not want to be named, put forward two situations in which the data could not be monitored: first, whether to bury the data (data burial is a key link in data collection), if the data is not buried, the data will definitely not be monitored; Second, it depends on whether the network architecture built by car companies can transmit data to the terminal in time, and if not, data will not be detected.

Experts call for vehicles to be equipped with "black boxes"

It is worth noting that in April this year, Tesla took the lead in launching the EDR suite query software in official website, USA. It is understood that EDR is an automobile event data recording system, which can be used to record the key data of automobile operation in three stages, including speed, ABS state, steering angle of steering wheel, airbag state and vehicle braking state.

Tesla said that the owner needs to spend $1,200 to purchase the cable and USB-to-CAN adapter needed for the connection, so that the vehicle can be connected, the driving data can be queried and the corresponding data report can be generated, which can be used to check the specific causes of the accident. "Tesla’s move respects the owner’s data control to a certain extent." He Yuan, executive director of the Data Law Research Center of Shanghai Jiaotong University, said publicly.

However, Tesla’s data query service is currently only open to North America, and Tesla users in China and other parts of the world are temporarily unable to use this function.

Image source: Every photo by reporter Liu Ling (data map)

It is worth noting that although Tesla has made it clear that the relevant domestic data query platform will be launched during the year, it has not explained the specific data query content provided by the platform and whether it is charged. "We will release the specific content of the data platform when it is launched in China." The relevant person in charge of Tesla China told reporters.

The reporter learned that domestic car companies, including Weilai, Tucki and Chang ‘an, have not yet built a platform for car owners to query vehicle data. "With the increasing intelligence of vehicles, domestic car companies can only provide some basic data for car owners. Users can’t view all the data unless a third party, such as the public security department, intervenes. " A person in charge of an independent brand car company who asked not to be named told the reporter.

At present, there are no relevant laws and regulations on the uploading of intelligent networked automobile data and the publication of data flow specifications by automobile enterprises after accidents in China. "It is unrealistic for car companies to release all the data, which involves the personal privacy of users." Cui Dongshu, secretary-general of the National Passenger Car Market Information Association, believes that car companies can classify the data and make clear what can be provided to car owners, and the relevant regulatory authorities should also make clear the scope of data provided by car companies.

Therefore, Song Jian, a professor in the Department of Automotive Engineering of Tsinghua University, suggested that the relevant authorities can delimit the vehicle data range that needs to be supervised and put it into the vehicle driving recorder. For example, important data such as steering, braking and acceleration are recorded in real time by the driving recorder, just like the "black box" of an airplane. "In the event of an accident, manufacturers and individuals have no right to dismantle the’ black box’, and only the national authority can complete it." Song Jian said.

Data integrity is as important as authenticity.

"The data types of autonomous driving are very complicated, including external environment data, vehicle data, user driving behavior and privacy data." The relevant person in charge of the China Automobile Association said.

In view of the network security of intelligent networked vehicles, the "Guidelines for the Access Management of Intelligent Networked Automobile Manufacturers and Products" (draft for comments) issued by the Ministry of Industry and Information Technology in April this year clearly requires that intelligent networked automobile products should have the functions of event data recording and automatic driving data storage, so as to ensure the integrity of equipment recording data in case of vehicle accidents, including driving automation system running status, driver status, driving environment information, vehicle control information, etc.

Image source: photo network

However, while ensuring the integrity of vehicle data storage, how to ensure the authenticity and tamper resistance of vehicle data is another important challenge facing the development of automobile industry.

In fact, Tesla emphasized the authenticity of the data more than once. "Tesla’s vehicle data is that the vehicle gateway reads the signals of various components in the car and stores them in encrypted form." Tesla said that the stored data is recorded by encryption technology, and it is impossible to directly read, modify or delete related data.

Even so, it has not eliminated the doubts about the authenticity of the data provided by the outside world. In order to solve this problem, China Automobile Association released the blockchain platform for trusted storage of automobile data at the end of April, which provides enterprises with trusted storage of data by using the unique characteristics of data tamper-proof and traceability of blockchain technology.

The China Automobile Association suggested that the government should revise and supplement the regulations and standards needed to adapt to the development of intelligent networked vehicles according to the different data types involved in intelligent networked vehicles in the future, and also suggested that the data supervision system of intelligent networked vehicles should be further improved by adopting a multi-centralized data governance model.

Reporter |Li Xing?Edit |Sun Lei Cheng Peng?Xiao Yong

Proofread |He xiaotao

Cover image source: national business daily Zheng Derui photo

| national business daily ?nbdnews??Original article |

Reprinting, editing, copying and mirroring are prohibited without permission.

If you need to reprint, please apply to this WeChat official account backstage and get authorization.