Motion Planning And Control For Autonomous Vehicle Driving On Motorways

The increasing traffic density of vehicle needs every motorist to keep high attentive-ness, any inattention may cause problems like automobile accidents and traffic jams which may endanger people's safety in the vehicle. Recent trends in automotive industry point in the direction of increased content of electronics, computers and controls with emphasis on the improved functionality and overall system robustness. Especially the early works on active safety systems date back to the 1980s, such as anti-lock brake systems (ABS), traction control systems (TC) and active front steering systems (AFS), are used to avoid accidents and at the same time facilitate better vehicle controllability and stability. These systems meet the vehicle stability through analyzing and controlling actuators. That is, drivers provide the vehicle with drive strategy and the Intelligent Transportation Systems are only stability controllers which compensate the wrong or critical driving strategies. Such a vehicle is a automotive vehicle in a sense, but it is automotive only on action, not on making decision. Obviously, the method of compensating wrong strategy can improve vehicle stability, but its control capability is limited. Therefore there is growing attention to the planning and active control for automotive vehicle. This thesis mainly studies the autonomous driving in highway for the autonomous vehicles.Firstly, a kinematic vehicle model is presented. Analyzing the characteristics of simplified vehicle kinematic model and general model, we present a planar dynamic vehicle model adapted to analyzing planar vehicle dynamics characteristics. We also discuss the methods of vehicle path planning, trajectory planning and dynamic control. However, these conclusions can't be validated on actual vehicle. So, this contribution needs a high-level degree vehicle model which can adequately reflect the main movements. For the reason of showing the effectiveness of proposed algorithm, this paper establishes a 14 degrees of freedom vehicle model, which describes the course of dynamic modeling for sprung and unsprung vehicles respectively in detail, and couples the suspension model with vehicle body model through force analysis of point connecting suspension and body. The coupled model reduces algebraic loops and improves the simulation speed. Then we combine vehicle transmission model with steering model and brake model, and establish a 14 degrees of freedom vehicle model in the software of Simulink.With the usual kinematic model, we discuss the trajectory planning problem for given reference path function. On the basis of introducing differential flatness and vehicle dynamics characteristics, this paper presents the course of usual form of dynamic model. To satisfy the lateral stability, we take the lateral acceleration into account when we plan the trajectory. The trajectory planning off-line method is fit for settled path cruise whose main goal is fuel economy. Therefore, based on the analysis of motors characteristic curve, we fit the curve and get the velocity and fuel function, and make it into a goal function when planning trajectory. Simulation results show that the trajectory planning method based on differential flatness reduces optimization dimensions. At the same time, the trajectory planning method based on differential flatness guarantees the stability and economy.At present, most automotive vehicles recognize road information through machine vision, and the result is the form of points. In order to obtain the path function, we need to fit the road information points, which doubtlessly increases computation burden and reduces real time of vehicle planning. Therefore, this paper proposes path following method based on path preview points. Due to the environment of vehicle is a structure environment which helps us make decision, we describe and define the structure environ-ment with mathematical language. Also, we define the availability and stability of lane. Then we present the MPC method based on differential flatness. For different operating environments, we propose corresponding MPC path following algorithm. When planning and controlling vehicle motion, we take the longitudinal acceleration and lateral accel-eration into account and make them as the constraints of MPC control method. The simulation results on 14 degrees of freedom vehicle model show availability and stability of our algorithm.Whether the automotive vehicles or the vehicles operated by drivers are afraid of the tire flat on the highway. The delayed reaction and operated mistakes of the drivers will lead to serious traffic accident. Similarly, if there isn't a controller designed based on tire characteristics and kinetic characteristic of the flat tire vehicle in the automotive vehicle, the designed controller only based on gross dynamics can't ensure the stability of the flat tire vehicle. For this reason, we describe the influence of tire performance changes on the vehicle stability under the condition of vehicle tire flat. Through the simulation and analysis of the different operated behaviors under the condition of tire flat, we present an ideal driving behavior and safety evaluation under this condition, and propose a planning control method under tire flat based on MPC. Compared with the conservative operation of driver under the condition of tire flat, our algorithm can promote the stability of the vehicle obviously under the condition of tire flat.In this paper, we not only have a series of clear demonstrations and simulations for the working conditions and the methods, but also give the deducing process of the vehicle planning and the controller design in detail. To validate the effectiveness of the proposed methods, we design some simulation experiments under typical and joint working conditions. In this paper, we have a series of detailed analysis and discussion of the vehicle system modeling, the trajectory planning based on flatness, the MPC method based on flatness, the description of the definition of structured environment and the characteristics of the vehicle tire burst. The results show that our planning methods and control methods for automotive vehicle on the highway are satisfactory.There are some topics that deserve further studying, such as ignoring the influence of the lateral wind on the characteristics of vehicle when we establish the 14 degrees of freedom vehicle dynamic model, so there will be modeling error during high speed turning. In addition, the process which describes the flatness output of the five degrees of freedom vehicle dynamics model is worth researching and discussing in the future.