A 10,000-word article on automobile control in the intelligent era

Abstract Autonomous driving is an important milestone in the development of the automobile industry. Automobile driving automation has been ongoing, and its development process is the gradual enhancement or eventual replacement of various important links such as driver cognitive perception, decision-making planning and execution control. In the intelligent era, disruptive technologies such as big data analysis, ubiquitous computing, ubiquitous sensing and artificial intelligence provide new opportunities for automobile driving automation to move towards a high level. Control technology is the cornerstone of automobile automation in the intelligent era. More information will derive more new functions and new systems under the empowerment of advanced control technology, thereby achieving improvements in various aspects such as automobile safety, economy and comfort. This article reviews automobile control in the intelligent era. First, it reviews the development process of automobile automation, then discusses the problems faced in the process of automobile automation, and finally sorts out some future development trends and key technologies of intelligent automobile control.

Automobile driving automation covers multiple industries such as automobiles, transportation, the Internet, electronics, and surveying and mapping. It integrates 5G communications, artificial intelligence, the Internet of Things, advanced robots, big data, cloud computing, intelligent manufacturing and many other disruptive technologies that will have a significant impact on future science and technology and human society. Automobile driving automation is an important stage in the development of the automobile industry. It has become the focus of attention of the global automobile industry. Both the government and enterprises are vigorously promoting the development and application of automobile automation technology.

In the J3016 automated driving classification, the Society of Automotive Engineers (SAE) of the United States divides automated driving vehicles into six levels, namely L0 to L5, according to the degree of automation, as shown in Figure 1, namely L0 full driver control, L1 to L2 driver assistance and partial automation, L3 conditional automation, L4 high automation and L5 full automation. In the Automated Driving Roadmap, the European Road Transport Research Advisory Council (ERTRAC) further discussed different types of vehicles based on the SAE automation classification, such as planning the development of passenger cars at all levels of automation from the perspective of functions and automated driving scenarios. my country has also given stage goals for the development of automated driving, and the research and development and industrial development plan for automated driving has been put on the agenda [1]. In addition, relevant policies such as road test environment and information acquisition (such as map surveying and mapping) involved in the development of automated driving are also being gradually relaxed [2−3].

From the perspective of car driving tasks, whether it is a new energy vehicle or a traditional internal combustion engine vehicle, a manned vehicle or an autonomous vehicle, the main task of car driving has not changed, that is, to achieve driving from the starting point to the end point, which is reflected in the operation level mainly for steering, accelerator, braking and gear operations. Figure 2 is a schematic diagram of car driving control. Human drivers are the most intelligent driving controllers. When the car is driving, people perform braking, driving and steering operations. The vehicle's execution system will calculate the corresponding vehicle longitudinal, lateral and vertical motion control signals based on these driving operation intentions and combined with the vehicle and traffic road conditions to achieve energy saving. The car can drive freely under the driving requirements of energy, safety and comfort. During this process, the human body can feel the vibration, inertia and reaction force of the steering wheel of the vehicle. The driver's senses such as eyes, ears and brain can fully perceive and understand the surrounding driving environment, and then make corresponding driving decisions based on driving experience and environmental cognition. The automation of automobile driving is actually the use of advanced sensing and autonomous interaction to achieve automatic operations such as steering, driving and braking on the premise of completing driving tasks, with the goal of improving performance indicators such as safety, energy saving, environmental protection and comfort. From the perspective of control, this kind of automation at the operational level is an enhancement or complete replacement of each link of the perception, cognition, decision-making planning and execution control of the closed-loop system of human-vehicle-road-environment.

In the development of automobile driving automation, automation at the perception level is to enhance and replace the perception functions of the driver's eyes, ears, and brain, and then assist or replace the driver's understanding of the environment, such as adding on-board perception cameras and radars, adding V2X, GPS, etc.; automation at the decision-making and planning level is to enhance and replace the decision-making and planning functions of the driver's brain. For example, in adaptive cruise control, the vehicle's assisted driving controller identifies the position, speed, and acceleration of the vehicle in front or obstacles, and implements driving and braking decisions and planning. Other assisted driving functions such as automatic parking, automatic overtaking, and other automatic driving in some working conditions all replace or enhance the driver's decision in specific scenarios; the enhancement at the execution level is mainly reflected in the improvement of the functions and performance of the steering, braking, driving, and shifting control systems.

Fig.1 Levels of automotive automation

Fig.2 Diagram of automated vehicle control system

The degree of enhancement and substitution of driver perception, decision-making, and execution is actually the basis for the classification of driving automation. As shown in Figure 1, the higher the degree of substitution and enhancement, the higher the degree of automation. It can be seen that this development did not appear suddenly, but has been going on with the development of automobile technology. In the early stage of driving automation, due to the limitations of sensing technology, control and computing chip performance, traditional automobile automatic control technology mainly focused on the execution level. For example, the automatic transmission realizes autonomous decision-making and execution of gear shifting during longitudinal driving, which completely releases the left foot and reduces the operating burden of the right hand; the adaptive cruise system can realize autonomous decision-making and execution of acceleration and deceleration within a limited range, releasing the right foot; the emergence of technologies such as electronic stability controller (ESC), emergency braking (AEB), and lane keeping assistant (Lane keeping assistant, LKA) all play a good auxiliary and enhanced role in the driver's operation. In the intelligent era, with the rapid development of ubiquitous sensing and ubiquitous computing, and enabled by advanced control technology, driving automation will gradually penetrate into various links such as perception, planning, and decision-making, thereby realizing the improvement of various aspects of automobile safety, economy, environmental protection, and comfort.

As a follow-up to the automotive control review article [4], this article aims to review the progress of automotive automation in the past 5 to 10 years, based on traditional automotive control. First, the development of automotive control is briefly reviewed from the perspective of longitudinal and lateral driving automation. Then, the main problems faced in the process of automotive automation are summarized in terms of perception, decision-making, and collaborative control. Finally, the development trend and key technologies of automotive control in the intelligent era characterized by big data and information fusion are discussed. It is hoped that this will provide some inspiration for readers engaged in automotive control in their future research directions. Due to space limitations, it is difficult to cover all aspects of the review, and any incomplete content will be continuously supplemented in future work.

1 Review of the development of automobile control

Since the review article [4] has discussed traditional automobile control in detail, this article will only briefly review it and make appropriate supplements to the latest developments. Vehicle motion is divided into longitudinal, lateral and vertical motion caused by uneven road surface. This article will review the current development status of automobile control systems from three aspects: longitudinal, lateral and lateral-longitudinal coupling. Since the vertical motion of the vehicle is mainly related to ride comfort, it will not be discussed in depth.

1.1 Review of the Development of Automobile Longitudinal Motion Control

The longitudinal motion of a vehicle describes the longitudinal driving characteristics of the vehicle, involving the power system, transmission system and braking system. In terms of the power system, the engine of a traditional internal combustion engine vehicle is the most important actuator in the longitudinal direction, and its control performance determines the vehicle's fuel economy, emissions and longitudinal driving quality. The successful application of the electronic throttle [5-6] officially opened the curtain of the era of engine automatic control, followed by key electronic control technologies such as air-fuel ratio control [7], ignition timing control [8], and idle speed control [9] that have been gradually applied. In response to the problems of complex engine systems and conflicting control requirements, Bosch proposed a powertrain control solution centered on torque demand in 1998 [10], which has now become the industry standard for powertrain control. With the increasingly stringent energy-saving and emission reduction regulations, engine advanced combustion control [11], complex intake control [12-13] and emission after-treatment control [1 4] has also been widely studied. For a detailed review of engine control, please refer to reference [15]. The transmission system is another important actuator based on the engine system to achieve vehicle longitudinal power transmission, improve transmission efficiency and fuel economy. With the development of electronic control technology, transmissions have evolved into various types (including AMT, AT, CVT, DCT and EVT, etc.), realizing the automation of the shifting process [16-18], thereby completely freeing the driver from manual shifting operations. With the development of electric vehicles, multi-gear shifting control for electric vehicles has become a research hotspot. References [19-20] studied the model-based shifting process and shifting rules based on the motor working efficiency MAP. With the increase in the degree of electrification, multi-energy source hybrid vehicle energy management technology [21] and pure electric vehicle wheel hub motor torque distribution [22] have also been research hotspots in recent years.