基于激光雷达的无人驾驶汽车道路、交通标志与障碍物识别方法

Recognition methods of road,traffic sign and obstacle for unmanned cars based on laser radar

在无人驾驶汽车技术中,现有的传统雷达容易受扫描角度和距离及复杂环境的影响,探测精度较低,同时使用传统的几何轮廓特征和运动状态特征的方法,无人驾驶汽车对环境的感知能力特别是障碍物的识别率和准确性也较低。针对这些不足,文章首先从道路、交通标志方面提出一种视觉感知方法,以及一种基于激光雷达的时空特征向量的障碍物识别方法来提高无人驾驶汽车对环境特别是障碍物的识别率和准确性。

In unmanned vehicle technology, traditional radar is easily affected by the scanning angle, distance and complex environment, detection accuracy is low, at the same time, when using methods with geometric contour and motion characteristics, the perception ability of unmanned vehicles to environment, especially the recognition rate and accuracy of obstacle is also low. In order to solve these problems, this paper puts forward a method of visual perception, and an obstacle recognition method of spatio temporal feature vectors based on laser radar from the road and traffic signs, to improve the unmanned cars on the environment, especially the obstacle recognition rate and accuracy.