Table of Contents
Summary................................................ ................................................................. ........I
Abstract ............................................. ................................................................. ............ II
Table of Contents ........................................ ................................................................. ..................IVChapter
1 Introduction........................ ................................................................. ............. 1
1.1 Research background and significance........................ ................................................................. ............. 1
1.2 Research status at home and abroad ............................. ................................................................. ......... 2
1.3 Current status of research on overtaking control methods for autonomous vehicles ............................. .............................5
1.4 Research content and technical route...................................... ................................9Chapter
2 Research on the decision-making mechanism of overtaking behavior of autonomous vehicles .................. 12
2.1 System framework of autonomous vehicles based on highway environment..... .................................. 12
2.2 Decision-making framework for autonomous driving overtaking behavior...... ................................................................. ........ 14
2.3 Overtaking behavior modeling based on finite state machine ............................. .................. 16
2.4 Summary of this chapter............ ................................................................. .................................. 19Chapter
3 Overtaking Trajectory Planning for Autonomous Vehicles.... .................................................. 20
3.1 Overtaking behavior characteristics Research................................................. ............................. 20
3.2 Safety distance for lane changing of autonomous vehicles........................................ ........................ 21
3.3 Vehicle lane-changing motion trajectory planning ............. ................................................................. ........ 25
3.4 Overtaking motion trajectory planning ............................. ................................................................. 34
3.5 Summary of this chapter................................................ ................................................................. .... 37
Chapter 4 Lane Changing and Overtaking Trajectory Tracking of Autonomous Driving Vehicles ............................. ....... 38
4.1 Introduction to model predictive control ............................. ................................................. 38
4.2 Automatic Establishment of driving vehicle dynamics model................................ .................. 40
4.3 Trajectory tracker design based on model predictive control............. .............................42
4.4 Autonomous vehicle trajectory tracking simulation experiment................................ .................. 46
4.5 Summary of this chapter ............................. ................................................................. ............. 52
Chapter 5 Verification of overtaking method for autonomous vehicles ............. ............................. 53
5.1 Construction of self-driving car overtaking model............. ................................................................. ....... 53
5.2 Analysis of overtaking results ............................. ................................................................. .. 56
5.3 Summary of this chapter........................................ ................................................................. ........ 59
Chapter 6 Conclusion........................................ .................................. 60
6.1 Research summary... ................................................................. .................................................. 60
6.2 Research Outlook................................................ ................................................................. . 61
Acknowledgments................................................ ................................................................. ........ 62
References ............................................. ................................................................. ........ 63
Obtain scientific research results related to the thesis during the degree study...................... ... 67
Chapter 1 Introduction
1.1 Research background and significance
The development of automobiles changes people's lives. Since the German engineer Carl Benz invented the first gasoline engine car in 1885, and the American Ford Company mass-produced cars in 1913, the automobile industry has experienced more than a century of development and gradually formed today's refined production - realized on a modular universal platform. Cross-model and cross-level large-scale production, including from car body structure to car functional module division, standard design, personalized customization, flexible manufacturing, flexible assembly and agile production. The core competitiveness of the automobile manufacturing industry has evolved from chassis, tires, machinery, transmission, and body in the 19th century, to engines, energy emissions, electrical, and passive safety in the 20th century, to today's modularization, automotive electronics, active safety, and intelligent driving. . Against the background of the rapid development of industries such as the Internet and artificial intelligence, electronic information technology is challenging the separation of drivers from driving cars. Self-driving car technology can reduce safety accidents and relieve driver fatigue, and is expected to change the entire automobile industry and improve traffic conditions. .
Domestic and foreign research institutions began research on autonomous driving technology as early as the last century. Since the 1970s, developed countries such as the United States, Britain, and Germany have begun research and development of autonomous driving. In recent years, major companies have rushed to participate in the research of autonomous driving technology and have successively carried out road tests to promote the rapid development of autonomous driving technology. my country's autonomous driving technology research and development is a little late. In 1992, the National University of Defense Technology successfully developed China's first truly driverless autonomous vehicle. In 2005, Shanghai Jiao Tong University successfully developed the first urban autonomous vehicle.
Self-driving car technology involves multidisciplinary interdisciplinary research such as artificial intelligence, vehicle engineering, automatic control, and machine vision. It uses multiple sensors such as cameras, radars, and navigation systems to sense the traffic environment where the vehicle is located, and automatically plans a safe and reliable driving trajectory, allowing self-driving cars to drive autonomously on the road. Overtaking is a common driving behavior. It means that in the same lane, the vehicle behind in order to maximize the speed, larger driving space and shorter driving time, overtakes the vehicle in front sideways and returns to the same lane. Lane behavior. As a relatively complex driving behavior, overtaking has huge safety risks. By collecting and analyzing a large amount of data, including vehicle speed, location, surrounding environment, etc., autonomous vehicle overtaking technology can better replace the driver in making safer overtaking decisions and controls, thereby reducing the risk of traffic accidents during vehicle overtaking. , it can also take into account the ride comfort of the vehicle and give passengers a more comfortable overtaking experience. Research on overtaking technology for autonomous vehicles is indispensable and of great significance in the process of achieving the goal of driverless driving.
1.2 Current status of research at home and abroad
1.2.1 Current status of foreign self-driving car research
In the 1980s, famous American universities and some scientific research institutions such as Carnegie Mellon University, Stanford University, and MIT began research on self-driving cars. In 1984, the U.S. Defense Advanced Research Project Agency (DARPA) released the "Star Wars" strategic plan, aiming to apply supercomputer technology and artificial intelligence technology to the military. Figure 1-1 shows the three DARPA The champion vehicles of the challenge are Sandstorm, Stanley, and Boss. At the same time, the United States Department of Transportation established the Automated Highway System (AHS) program, with General Motors, Berkeley University and Carnegie Mellon University participating in the project.
As a major participant in the AHS program, Carnegie Mellon University has developed the NavLab series of vehicles. Figure 1-2 shows the NavLab-5 autonomous vehicle. In 1995, the smart car NavLab-5 completed a 4,585-kilometer smart driving road test from Pittsburgh to San Diego. During the test, the vehicle autonomously controlled the steering wheel for approximately 98.2% of the total mileage. In 1987, Europe launched the PROMETHEUS (Programme for a European Traffic of Highest Efficiency and Unprecedented Safety). The project consists of the famous Bundeswehr University Munich and famous companies BMW and Mercedes-Benz as main participants. The project team developed intelligent driving vehicles VaMP and VITA-2 in 1994, and mixed them into the normal traffic flow of the highway. The maximum speed of the vehicle reached 130km/h. During the test, items such as line patrol, formation, tracking, lane changing and overtaking were demonstrated. Compared with NavLab-5, collaborative control of steering wheel, throttle and brake is added. In 1996, the ARGO project team was created by the Vision Laboratory of the University of Parma, Italy. The project uses computer vision to identify lane markings one by one and then control vehicle driving. Figure 1-3 shows the ARGO self-driving car. In 2010, the ARGO test vehicle autonomously drove to China to participate in the Shanghai World Expo along the Marco Polo route, with a total distance of 15,926 kilometers.
1.2.2 Current status of domestic self-driving car research
The first domestic institutions to start research in the field of self-driving cars were universities. On the one hand, colleges and universities are actively cooperating with automobile companies to clarify the path from laboratories to product industrialization; on the other hand, they are incubating related technologies and products within the colleges.
In the late 1980s, China's famous universities, including the National University of Defense Technology and Tsinghua University, jointly developed my country's first autonomous vehicle, the ATB-1 (Autonomous Test Bed). In the mid-1990s, Tsinghua University established a smart car research and development team. Professor Li Keqiang of Tsinghua University proposed that the development of smart cars is developing in two directions: intelligence and networking. The former uses vehicle-equipped sensors to sense the external environment and complete "isolated" autonomous driving, while the latter uses vehicle-to-vehicle communication and infrastructure information exchange. Realize
autonomous driving under network connection. Both development directions have the ultimate goal of liberating human hands, and will eventually be combined with each other to become "intelligent connected cars." In the process of cooperation with enterprises, the smart car R&D team is mainly responsible for providing systems and basic algorithm frameworks, and considering optimization around the specific details of the framework and the actual driving environment. Tongji University and SAIC Motor took the lead in establishing a collaborative innovation center for intelligent new energy vehicles, dedicated to helping commercialize projects. Tongji University conducts exploratory basic research on SAIC Group's product planning and industry trends, while SAIC Group's Foresight Department is committed to technology productization. For the interdisciplinary system engineering of intelligent connected cars, the Collaborative Innovation Center brings together teachers and students from various colleges, including the School of Automotive, School of Software, School of Telecommunications, School of Transportation, and School of Surveying and Mapping, to leverage their respective strengths to complete the project. In 2003, the National University of Defense Technology and FAW Group jointly developed the Hongqi CA7460 intelligent driving vehicle, which functionally realizes automatic overtaking. The second-generation intelligent driving vehicle HQ3 developed in 2006 has adaptive cruise, collision warning, lane tracking and other technologies, and controls The accuracy and stability have been improved compared to the first generation.
China's local automobile companies responded to the national call and began to participate in the research and development of autonomous driving technology. In April 2018, Chery released the "LION" intelligent brand, which is an upgrade of the "124" strategy and involves R&D, manufacturing, products, marketing, services, etc., including autonomous driving, intelligent interconnection, smart manufacturing, digital With these five basic points of marketing and mobility, we attempt to complete fully autonomous driving in four stages: Level 1 driving assistance in 2006, Level 2 partial autonomous driving in 2018, Level 3 conditional autonomous
driving /Level 5 fully autonomous driving. SAIC Motor proposed the "four new modernizations" of technology in its "13th Five-Year Plan": electrification, networking, intelligence and sharing, and gradually formed an independent research and development system for intelligent connected vehicles, laying the foundation for future product and business expansion. In 2015, SAIC stated that it would achieve autonomous driving on structured and some non-institutional roads within five years, and autonomous driving in all environments within 10 years. Its autonomous driving technology is based on Level 3 smart cars as the starting point, and promotes technological development around the two main lines of vehicle intelligence and multi-vehicle collaboration. In 2016, Changan Automobile formulated the "654 Strategy" and built six major platforms (electronic and electrical appliances platform, environment perception and execution platform, decision-making platform, software platform, environmental testing platform, standards and regulations platform) and five core technologies (automatic parking platform) for the intelligent sector. vehicle technology, adaptive cruise technology, intelligent network technology, HMI interactive technology) and realize autonomous driving in four stages.
As an Internet company, Baidu has also conducted systematic research on smart cars, covering areas such as Internet of Vehicles, high-precision maps, and the development of autonomous driving software and algorithms. Baidu's research on smart cars is divided into two branches: Internet of Vehicles and autonomous driving. In 2017, Baidu established an independent autonomous driving division (Level 4) and announced the "Apollo Project" for commercial open source autonomous driving.
1.3 Research status of overtaking control methods for autonomous vehicles
Vehicle overtaking is a common driving behavior, which refers to the driving behavior of the vehicle behind overtaking the vehicle in front in order to seek faster driving speed and larger driving space. When an autonomous vehicle overtakes, it determines whether it meets the overtaking conditions based on the surrounding environment, reasonably plans the overtaking trajectory, and then makes decisions based on the vehicle's current attitude and speed information, outputs a steering wheel angle signal, and completes the overtaking behavior. The overtaking process involves tasks such as lane-changing overtaking decision-making, lane-changing overtaking trajectory planning, and lane-changing overtaking trajectory tracking.
1.3.1 Current status of research on lane-changing and overtaking behavior decision-making
Early domestic and foreign research on driving behavior decision-making mostly stayed at the simulation stage, using behavioral logic to imitate real driver driving behavior habits. Gipps was the first to systematically study the lane-changing behavior of cars. The lane-changing decision model he proposed was based on suburban roads affected by obstacles, traffic signs, and heavy vehicles. The decision-making process was divided into generation of lane-changing intention, judgment of lane-changing conditions, The lane-changing action executes three parts. In order to reduce the complexity of the model, hierarchical decision-making is adopted to make the decision-making meet various requirements [2]. Hidas improved on
the model , proposed the SITRAS (Simulation of intelligent Transport Systems) model, and proposed a gap evaluation model to determine whether the acceleration and deceleration of the front and rear vehicles of the current vehicle are acceptable when judging the feasibility of lane changing, avoiding the risk of own vehicle Lane-changing behavior has a negative impact on other traffic vehicles [3]. Q.Yang proposed the MITSIM (Microscopic Traffic SIMulator) model based on the Gipps model framework. It was also the first to divide lane changing behavior into mandatory lane changing and non-mandatory lane changing according to different environments. Among them, non-mandatory lane changing The lane change intention is generated by adding the expected vehicle speed as an indicator. The US Federal Highway Administration proposed the CORSIM lane changing model, which uses two microscopic simulation models for different types of roads, namely the FRESIM model suitable for highway environments and the NETSIM model suitable for urban road environments. The FRESIM model consists of motivation factors, interest factors and emergency factors. The NETSIM model is divided into two lane changing situations: mandatory and non-mandatory. These two models are established to judge the lane changing timing based on deceleration.
Most of the above models simplify the environment and assume that all micro-environmental information is knowable, which is not consistent with the actual situation. Moreover, the above model only considers driving intention once, treats lane changing and overtaking behavior as a continuous action, and does not consider whether the surrounding environment meets the conditions when the overtaking behavior changes lanes for the second time. Therefore, more in-depth research is needed to build models that can be applied to autonomous vehicles. Schubert et al. used deceleration time as a decision-making index in the lane-changing process, and used Bayesian network to evaluate lane-changing scenarios and make lane-changing decisions. Wei et al. used a predictive model to analyze the dynamic driving environment and then assist vehicles in completing driving behavior decisions such as lane keeping or vehicle overtaking on the highway. The model uses vehicle driving safety, comfort, and efficiency as evaluation indicators, and uses a cost function to make decisions. Based on this, Markov methods are then used to improve the driving stability of the vehicle in uncertain driving environments. Brechtel et al. used the Markov method as a lane-changing decision-making method. The decision-making conditions used directly measurable physical quantities such as relative distance and relative speed. However, due to unpredictable system measurement errors, the decision-making system was unstable. In terms of corporate research, BMW's autonomous driving ConnectedDrive project is based on highway research and development. Ardelt and others use state machines to distinguish different driving behaviors and make hierarchical decisions. The hierarchical decision-making defines state transition conditions based on different driving subtasks and driving environments.
In China, many scientific research institutions have also conducted in-depth research on the overtaking behavior decision-making of autonomous vehicles. Yuan Shengyue conducted research on lane changing rules for urban environments [9]. Based on the division of drivable areas, Guo M et al. proposed a decision-making model suitable for autonomous vehicles. The decision-making model takes into account other information, including traffic lights, surrounding vehicles, pedestrians, etc. [10]. Xu Youzhi and others learned the driving characteristics of real drivers based on RBF neural network, obtained the conditions for overtaking intention generation and judgment, and built a simulation experiment platform based on Prescan and Matlab/Simulink to verify the effectiveness of the overtaking decision-making framework [11].
.....If you need the complete version of the material, send a private message