2.1

Auto-drive system

The so-called autonomous vehicle is an intelligent network driving process of driverless driving with the help of computer systems. The autonomous vehicle relies on the cooperation of artificial intelligence, visual computing, radar, monitoring equipment and global positioning system. The intelligent networked control system makes the vehicle and road perfectly fit without the intervention of the driver, and automatically and safely drives the motor car ⁵.

Automobile manufacturers and technology companies face huge obstacles when deploying autonomous vehicle, because safety standards formulated decades ago require manual control of auto-drive system. According to the old regulations of NHTSA, the vehicle “must have a driver’s seat, a steering wheel and steering lever, or only one front outboard passenger seat”. However, NHTSA stated that “vehicles completely controlled by the auto-drive system do not need manual driving control.” Therefore, NHTSA amended the regulation.

NHTSA said that the current regulations do not prohibit the deployment of autonomous vehicle as long as they are equipped with manual driving control systems. NHTSA will continue to consider changing other safety standards, but manufacturers may still need to apply to NHTSA for exemption before they can sell their vehicles equipped with auto-drive system. In recent years, the penetration rate of the new energy vehicle market has been increasing under the condition of new vehicles, which has led to the entry of many industry giants. For example, the entry of Internet companies such as Xiaomi, Baidu and Huawei has surged again in this wave of intelligence. Admittedly, the birth of each brand must have its inherent ambition and blueprint ambition. The founders of each brand are eager to make their cars more colorful and even change the industrial structure and the whole world. To fully realize automatic driving, we need to solve technical problems, that is, effectively integrate multiple sensors in the entire vehicle equipment: millimeter wave, laser radar and camera, and analyze and process the sensor data through computer; Secondly, we need to solve data and map problems ⁶. The map supporting unmanned driving is not two-dimensional abstract data, but requires high-resolution and high-precision data, so the amount of data stored and transmitted will be large. Once the data is connected to the system center of the vehicle, the intelligent automatic driving algorithm of AI technology can be used to identify the environment and realize intelligent automatic control ^7-8.

As one of the few automobile driving companies in China that can achieve the mass production and delivery of TDA4VM, Nullmax has completed the fixed point of integration of parking lot and parking lot since 2020 and 2021. Nullmax has chosen the integration of parking and parking from the scene to the automatic driving landing from the early stage of its business, and has created the integration of parking and parking for the entire scene.

At present, the field of integrated navigation and berthing is divided into several forms:

The independent system of two SOC and two domain controllers is used for driving and parking, which is called 1.0 form;

Integrate two SOCs into a domain controller to reduce costs through a small amount of integration, which is called 2.0 form;

A single SOC and integrated domain controller are used, but the sensor still does not have the 3.0 form of deep reuse.

Automatic driving technology is not only the modernization of traditional cars, but also a major project to upgrade the entire driving system, involving the upgrading of the car itself, the upgrading and improvement of traffic infrastructure, the complement and establishment of social infrastructure, the maturity and wide application of intelligent network technology, the promotion of laws and regulations, and even the change of the driver’s sense of responsibility and behavior habits ⁹. As the upgrade of the whole transportation industry, automatic driving will change people’s travel mode in the future. Its working principle mainly depends on the computer system inside the car and the sensors around the car body, and the information is processed remotely through the cloud server and fed back to the vehicle to guide the driving, so that the car can drive itself without driver intervention ¹⁰, as shown in Figure 1.

Figure 1.

Schematic diagram of the automatic driving system

According to the developed classification standard of the degree of automation, the degree of automatic driving of vehicles can be divided into several levels ¹¹, as shown in Table 1.

Table 1.

Degree of automatic driving of vehicles

Automatic driving technology cannot be separated from the support of big data and algorithms. In this work, we studied the traditional autonomous driving target detection module using SVM (support vector machine) method in intelligent control algorithm. In the process of practical application, SVM needs to use kernel function for feature transformation ¹².

The common classification of kernel function includes polynomial kernel function and Gaussian kernel function, and their corresponding forms are as follows:

Polynomial kernel function:

Where, a is the adjustment parameter and d is the maximum number of times.

Gaussian kernel function:

∥·∥ represents the module of the vector.

2.2

Collection, utilization, protection and compliance management of personal data in the vehicle

2.2.1

Collection and utilization of personal data in the vehicle

According to the purpose of collecting data from driverless vehicles, data can be divided into two categories: traffic and lifestyle. Comprehensive access to external information is crucial to the safe driving of driverless vehicles. In the traffic category, the data generated using V2X technology can be further classified ¹³.

For personal data in the car, a large amount of data needs to be mined, processed and shared. For example, for the requesting end, we need to request service from the remote end and send the current vehicle status; For the server, we need to collect a large number of vehicle and road information, and then extract, analyze and contextualize the value data to provide personalized services for the requester in real time.

The process of extracting potentially useful knowledge or patterns from incomplete, fuzzy, noisy, highly random and previously unknown real-world application data is called data mining. If you use the traditional method to simply operate directly with text, you need to traverse the text line by line. The small amount of data may be faster, but the processing efficiency of massive data is not as high as Hive.

Hive is the basis of Hadoop-based data warehouse and is a simple and easy-to-use parallel data processing tool. In the big data environment of vehicle network, the limitations of traditional methods are increasingly prominent. Therefore, the data to be collected and extracted in this article will be imported using the Hive mechanism, as shown in Figure 2 ¹⁴.

Figure 2.

Hive architecture diagram

3.1

Comparison and analysis of personal data collection in the vehicle

This paper collects data through Hive, and obtains personal data in the car in two ways: Hive data collection and traditional collection. This paper collects 1000 data, of which 998 points are valid and 2 points are invalid. The selection criteria are based on the data from the last five years.

This study compares and analyzes the collection speed of Hive data collection and traditional collection of personal data in the car, and the results are shown in Figure 3.

Figure 3.

Comparative analysis chart of in-vehicle personal data collection

From the data in Figure 3, we can see that the traditional data collection speed is: music preference 23 minutes; The temperature preference is 12 minutes; Use the light for 8 minutes; Place collection for 9 minutes; Common routes take 17 minutes. Hive data collection speed: 4 minutes for music preference; Temperature preference 1 minute; Use the light for 2 minutes; Place collection for 1 minute; Common routes take 3 minutes. It can be seen that Hive’s data collection speed is far faster than the traditional way in any aspect, which will provide effective technical support for the research of personal data in the car in the subsequent development of auto-drive system.

3.2

Comparative analysis of personal data protection

This study compares and analyzes the number of attacks and the number of successful defense of the personal data system in the vehicle in the new and traditional modes within a week. The results are shown in Figure 4.

Figure 4.

Comparative analysis chart of personal data protection

It can be seen from Figure 4 that the traditional data management system has been maliciously attacked 9 times and successfully defended 6 times in a week; The new data management system has been attacked twice and successfully defended twice at the same time. This shows that the use of a new data protection system can not only reduce the number of external attacks by about 33%, but also effectively defend attacks, with a 33% increase in the defense success rate. This is of great significance for the safety protection of personal data in the car during the development of the auto-drive system, and will escort the research of personal data in the car during the subsequent development of the auto-drive system.