摘要
随着科技发展,激光雷达在无人驾驶车中的应用成为社会热门的话题。其中,栅格地图也成为动态障碍物检测的手段之一。原始贝叶斯推理方法在栅格概率趋于极值时,若栅格状态发生改变则体现出的很强的滞后性,因此提出一种利用模糊逻辑矫正权值变量对贝叶斯后验概率进行限制的算法。应用改进的贝叶斯推理更新栅格状态并利用冲突变量检测动态障碍物。最后,通过膨胀、腐蚀、改进连通区域标记法及一维数据区间密度算法提取障碍物信息及可行驶区域信息。实车实验表明提出方法的有效性。
With the development of science and technology,the application of lidar in unmanned vehicles has become a hot topic in society. Among them,the grid map has become one of the means of obstacle detection. The original Bayesian interference method has a strong hysteresis once the grid state changes when the grid probability tends to be extreme. Therefore,a new method is proposed to estimate the Bayesian posterior probability based on fuzzy logic which can correct the weight variable. The improved Bayesian interference is used for updating the grid state and the conflicting variable is used for detecting dynamic obstacles. Finally,the obstacle information and the travelable area information are extracted by expansion,corrosion,improved connectivity region labeling method and one dimensional data interval density algorithm. The experimental results show a better effectiveness of the proposed method.
出处
《激光杂志》
北大核心
2017年第8期13-18,共6页
Laser Journal
基金
北京市属高等学校人才强教计划资助项目(038000543115025)
关键词
无人驾驶
栅格地图
模糊逻辑
贝叶斯推理
障碍物检测
unmanned vehicles
grid map
fuzzy logic
bayesian interference
obstacle detection
