基于信息融合的自动泊车系统车位线车位识别和决策规划

通过融合距离和视觉信息，提出了一种基于信息融合的车位线车位识别方法。首先利用摄像头采集车位角点在俯视图像中的坐标，然后采用基于逆透视变换的平面测距模型获取车身侧面距车位角点的侧向距离，最后融合超声波传感器和车辆里程信息建立车位线车位平面模型，确定了车辆后轴中心点坐标及相对初始偏航角，实现车位线车位的识别。根据识别出的车位信息进行路径规划并仿真分析，经过实车试验验证，该车位识别方法能够有效实现对车位线车位的识别，所规划路径合理、可行。

Parking Line Parking Slot Recognition and Decision Planning of Automatic Parking Systems Based on Information Fusion

A parking space recognition method was proposed based on information fusion by integrating distance and visual informations. Firstly, the coordinates of the corners of parking space in overhead image were captured by camera. Then the lateral distances of the vehicle body from the corners of the parking space were obtained by using the plane ranging model based on inverse perspective transformation. Finally, the parking line parking plane model was established by combining ultrasonic sensor and vehicle mileage information. The plane model determined the coordinates of the center point of rear axle of vehicle and the relative initial yaw angle to realize the recognition of the parking line parking slot. Based on the identified parking information, path planning and simulation analysis were carried out and verified by the real vehicle tests. Results show that the parking identification method of the vehicle may effectively realize the identification of parking line parking space, and the planned path is reasonable and feasible.