基于信息融合的自动泊车系统多工况车位识别和决策规划

Multi-condition Parking Space Recognition Based on Information Fusion and Decision Planning of Automatic Parking System

针对自动泊车系统中对斜式车位识别的缺失,在原有平行车位和垂直车位识别基础上,通过距离和视觉信息的相互融合,提出一种智能识别日常生活中斜式车位的方法,并能够完成多工况车位的识别及泊车。该方法利用超声波传感器、摄像头和车轮速度传感器的信息得出车辆的车身姿态特征参数,采用Sugeno型模糊推理方法计算出水平、垂直和斜式三种不同的车位类型;使用Matlab/simulink搭建多工况车位识别模型和相应路径规划模型,并针对六种不同类型斜式车位、平行车位以及垂直车位场景进行仿真分析,并进行实车试验。仿真和实车试验验证上述车位识别方法及路径规划策略的有效性和合理性。

Aiming at the lack of recognition of inclined parking spaces in automatic parking system,a method of intelligent recognition of inclined parking spaces in daily life is proposed,and multi-condition parking spaces recognition and parking are achieved.Based on the recognition of existing parallel and vertical parking spaces,the method is combined with distance and visual information.In this method,the information of ultrasonic sensor,camera and wheel speed sensor are used to obtain the pose feature parameters of a vehicle,the Sugeno fuzzy inference method is used to to determine three different types of parking spaces:horizontal,vertical and inclined.subsequently,the multi-condition parking space identification model and the corresponding path planning model are built in Matlab/Simulink soft,and the simulation is carried out for six different types of inclined parking spaces,parallel parking spaces and vertical parking spaces.finally,the real vehicle test is carried out.The simulation and real vehicle test results show the effectiveness and rationality of the above parking space identification method and path planning strategy.