基于多激光雷达与组合特征的非结构化环境负障碍物检测

发布时间：2018-01-05 20:40

本文关键词：基于多激光雷达与组合特征的非结构化环境负障碍物检测　出处：《机器人》2017年05期 　论文类型：期刊论文

【摘要】：针对非结构化环境下自主式地面车辆(ALV)的负障碍物检测问题,提出一种基于多激光雷达与组合特征的方法.首先,设计了一种具有互补能力的多激光雷达安装方式.其次,提出了基于幅向局部凸性和后沿壁局部密集特征的64线雷达负障碍物特征点对检测方法,以及基于径向距离跳变和后沿壁局部密集特征的32线雷达负障碍物特征点对检测方法.进而从负障碍物的时空融合角度,采用贝叶斯法则对多传感器多帧特征点对进行融合,然后采用DBSCAN(density-based spatial clustering of applications with noise)对融合后的特征点对进行聚类与过滤.最后对数据进行栅格化,提取负障碍物栅格.实验结果表明,本方法在非结构化复杂地形环境下具有良好的负障碍物检测性能.
[Abstract]:Aiming at the negative obstacle detection problem of autonomous ground vehicle (ALV) in unstructured environment, a method based on multi-lidar and combined features is proposed. A complementary multi-lidar installation method is designed. Secondly, a method for detecting the negative obstacle feature pairs of 64-line radar based on the local convexity of amplitude and the local density of the back wall is proposed. And the detection method of 32 lines radar negative obstacle feature point pair based on radial distance jump and local dense feature of the back wall, and then from the angle of space-time fusion of negative obstacle. Bayesian rule is used to fuse multi-sensor multi-frame feature point pairs. Then use DBSCANs (. Density-based spatial clustering of applications with. Clustering and filtering are carried out on the fused feature pairs. Finally, the data is rasterized. The experimental results show that this method has good negative obstacle detection performance in unstructured complex terrain environment.
【作者单位】：南京理工大学智能机器人研究所;
【基金】：国家自然科学基金(61473154,61371040,61373063,91420201) 国家装备预研领域基金(6140312010101)
【分类号】：TN958.98;TP242
【正文快照】： 1引言(Introduction)环境感知技术是ALV在室外环境中实现自主导航的关键技术.非结构化环境[1]中存在着陷坑、壕沟等低于地面且严重影响ALV安全行驶的负障碍物.因此准确可靠的负障碍物检测成为在非结构化环境下ALV环境感知技术的重要组成部分.Matthies等人[2]提出基于热红外图

【相似文献】