Research On Decision Control System For Automatic Parking

Prompted by the development of vehicle industry,enormous urban traffic pressure arises n most situations based on the increasing of car ownership,and the difficulty of parking has become one of the terrible problems for most of driver.Self-parking system can completely help drivers in parking operation.Self-parking decision control system is one of the most cimportant decision control system of the automatic parking system.The main idea of this thesis is to present a decision control system for self-parking in the field of automatic parking decision control system.This thesis presents a joint algorithm of modified Bellman-Ford algorithm and bidirectional breadth-first search algorithm to generate shortest parking path for vehicle.The simulation and experimental verification are put forward by this thesis.The main research of this thesis can be listed as followed:Firstly,according to the characteristic of parking path planning algorithm,vehicle kinematics modeling,and the map information of the conclusion of mature environment perception algorithms,the space constraints is generated for generating the grid map.The result of generating grid map is the basis of generating weighted ring graph.All these processes are the initializer of the decision control system for self-parking.Secondly,the parking path is generated by using bidirectional breadth-first search algorithm.The research analyzes the program run time and performance based on different size of grid map.The conclusion of this idea shows a phenomenon that shortest path can not be obtained by using bidirectional breadth-first search.algorithm alone.Thirdly,the thesis presents an modified algorithm based on queue optimization Bellman-Ford algorithm,and this algorithm can optimize the path planning conclusion which is generated by bidirectional breadth-first search algorithm.Meanwhile,the thesis compares the difference of performances between using bidirectional breadth-first search algorithm alone and joint algorithm.The thesis analyzes the conclusion,and obtains final shortest parking path.Fourthly,the thesis uses cubic spline interpolation algorithm to smooth the parking path.During the whole process of real parking,the vehicle direction is always in a smooth continuous state of change.Furthermore,according to the real parking path,the thesis completes the simulation of parking path.Fifthly,the research test platform is built by this study.According to the situation of using open sources of mature three-segment parking algorithm in this test platform,the experiment compares the performance of parking decision made by this thesis to the performance of traditional three-segment parking algorithm.After using two separate parking path decision control system in different conditions of parking environment in the particular one vehicle test platform,and the results of comparative experiment,the thesis presents the final conclusion of the whole algorithm.