基于视觉的车位识别与智能小车平行泊车

Parking Space Recognition and Parallel Parking of Intelligent Cars Based on Visual

文章针对自主平行泊车过程中,在不同场景下因环境因素导致难以识别车位的问题,提出了一种基于视觉的车位识别方法,并设计了相应的车位图像处理的算法。在此基础上,还提出了一种固定转弯半径的逆向泊车路径规划方法,该方法结合了两圆弧相切法。以具备独立转向功能的四轮智能小车为研究对象,以评估在4种不同车位空闲状态的场景中车位实时识别能力和自主平行泊车功能为任务,进行实车试验研究。试验结果表明:通过摄像头实时采集车位视频信息,运用车位图像处理算法以及路径规划方法,成功实现了车位准确识别及安全平稳泊车,证明了所提出方法的可行性。

In addressing the challenge of difficulty in recognizing parking spaces due to environmental variables during the process of autonomous parallel parking,a vision-based parking space recognition method was proposed,along with the corresponding algorithm for processing parking space images.Additionally,a reverse parking path planning method with a fixed turning radius was also introduced,integrating the two-circle tangent method.A four-wheeled intelligent vehicle equipped with independent steering capabilities served as the subject,tasked with evaluating real-time parking space recognition and autonomous parallel parking functionality in four different scenarios of parking space availability.This was accomplished through practical vehicle experiments.The experimental results show that the parking space information video was collected by the camera in real time,and the parking space image processing algorithm and path planning method are used to successfully realize the accurate parking space recognition and safe and stable parking,which proves the feasibility of the proposed method.