Research On Lane Tracking Control For Intelligent Vehicle

The intelligent vehicle, which has integrated a variety of high-techs, is not only the development direction of future automobiles, but also the updates of traditional automotive industry. In recent years, lane keeping system, as one of the key technologies of the intelligent vehicle, has been extensively studied. This system is designed for controlling the vehicle to drive according to the desired driving lane, or reducing the lane departure traffic accidents caused by negligence and fatigue driving. In this paper, the lane keeping system for intelligent vehicles is studied, a lane model is established based on the image information while the expected path is programmed. Besides, a constrained model predictive controller is designed based on the vehicle dynamics and kinematics characteristics to realize the lane tracking control.In this paper, the lane model based on image information is obtained by coordinate transformation. Then the description of the front lane curvature, the lateral displacement deviation relativly to the lane and the yaw angle deviation of the vehicle is presented. Besides, the method of on-line identification of lane model parameters based on discrete lane data points provided by video processing program of on-board camera is discussed and the feasibility and effectiveness of the lane model are analyzed through the experimental data.In order to solve the lane tracking problem in lane keeping system, a rolling horizon predictive control method is presented in this paper. A bicycle model with two degrees of freedom is established to describe the vehicle dynamics, and the expected path planning method based on the lane model is discussed. Considering the constraints of wheel angle and horizontal yaw rate constraints during the driving process, a constraint model predictive control method which can guarantee stability is designed to track the desired path. Moreover, the control performance is simulated and analyzed in the high precision vehicle dynamics simulation software ve DYNA while the feasibility and effectiveness of the control method are tested.In order to meet the implementation demand of vehicle embedded system, high performance visual development platform Jacinto6 and real-time system d SPACE are employed in this paper, and the lane model module and controller module of the embedded system are realized, respectively. The lane model identification data flow based on the image information and the data stream information transmission method based on CAN bus are established on Jacinto6. For the purpose of improving the real-time performance, the explicit predictive control method is adopted in the embedded system. The feasibility and applicability of explicit predictive control are analyzed through the comparison of control performances between explicit and optimization predictive control under the same simulation conditions.