Research On Trajectory Planning Of Auto_Parking Based On Differential Flatness And Spline Theory

Autonomous-parking technology is a hot research topic in recent Intelligent Vehicle application research area, and the trajectory planning is one of its key technology. So trajectory planning based on differential flatness and spline theory is mainly researched in this thesis, the flatness attribute of kinematics and dynamics of car model is investigated and the problem of parameterization of flatness output is good worked out. With the method of this thesis applying in the Autonomous-Parking system successfully, the trajectory planning problem of Autonomous-Parking could be solved effectively.The autonomous vehicle system is a nonlinear system restrained by the nonholonomic constraints, so the trajectory planning of vehicle has to fulfil the nonholonomic constraints, dynamic constraints and etc. As the attribute of a system, the differential flatness could map the inputs and states of a system into a output space, so the trajectory planning problem which is restrained by the nonholonomic constraints would be solved well. Trajectory planing in output space could reduce the difficulty of problem solving and more important is that the trajectory generated could be followed. The main research results of this thesis are as follows:The recent research work about Auto_Parking system is summarized and the connection between control and planning is analyzed. Then the difference between trajectory planning and path planning is elucidated and the difficulty of trajectory planning of the Auto_Parking system is investigated.The definition and property of the differential flatness are presented in this thesis. A trajectory planning approach based on differential flatness and spline theory is proposed and particularly how to parameterize the flat output using B-spline is explained. Experiments show that the demand of nonholonomic constraints and dynamic constraints could be satisfied.Considering the kinematics model and dynamic model of vehicle, direct method and dynamic feedback linearization are chose to determine the flat attribute of these two models and the flat outputs are found.After analyzing the Auto_Parking environment, a collision free space generation method of parallel parking is proposed. In this space, the method based on differential flatness and spline theory is applied and a feasible parking trajectory which satisfy the kinodynamic constraints could be generated.Experiments indicate that the generated trajectory using the proposed method would not only avoid the obstacles effectively but also satisfy the constraints and overcome the problem of steering at a fixed point. So the trajectory planning problem of Auto_Parking could be solved effectively.