摘要
提出一种优化地面分割与回环检测策略的激光雷达同时定位与建图(SLAM)算法,首先,建立同心区模型,使用主成分分析(PCA)算法来提取点云区域性统计特征,设计地面似然估计二分类方法去除地面上可能出现的非地面点。其次,在特征中提取模块,通过平坦度与球度系数来额外提取球体特征,以此来增加提取点云帧上的关键点。最后,优化回环匹配策略,采用基于马氏距离的鲁棒解耦的全局准配策略,矫正累积误差,提高定位与建图精度;在公开数据集、实际环境中评估算法性能与验证实际应用偏好,并与LOAM、LeGO-LOAM、FAST-LIO等算法进行对比。结果表明,相比于LeGO-LOAM算法,所提算法在定位精度和稳定性上均有大幅度提升,其中在具有回环序列00和02上定位精度提升分别达到82.92%和83.38%,在无回环序列10上定位精度提升达到63.18%,现场验证表明所提算法满足实际应用需求。
Objective As a crucial aspect of automatic driving,unmanned vehicle technology has garnered extensive attention and research in both academia and industry.Autonomous vehicles require robust perception and decisionmaking systems for autonomous navigation,with simultaneous localization and mapping(SLAM)being one of the core components.While many advanced SLAM algorithms have achieved stable and highprecision positioning and mapping,challenges persist.For example,nonsmooth and uneven roads can distort collected data,making it difficult to establish reliable feature correspondence between frames,leading to significant map drift and positioning errors.Given the inaccuracy of existing laser SLAM algorithms in ground segmentation,low featurematching efficiency,and the high computational demands of traditional loop closure detection methods based on Euclidean distance,we propose a lidar SLAM algorithm with optimized ground segmentation and closedloop detection strategy.Methods We first introduce a more reliable ground segmentation method for nonsmooth roads during the preprocessing stage.By establishing a concentric region model for each point cloud frame and using principal component analysis(PCA)to extract statistical characteristics,we design a ground likelihood estimation binary classification method to remove unstable ground points.This approach addresses the issue of misclassifying nonground points as ground due to small slopes between adjacent laser points,achieving more accurate segmentation of nonsmooth road surfaces.Additionally,we introduce a sphere feature point extraction method alongside the standard edge and plane feature points.This enhances the point cloud’s useful information for matching between consecutive frames,improving both efficiency and robustness while reducing the influence of redundant point clouds.In addition,we propose a robust global alignment strategy based on Mahalanobis distance to replace the traditional iterative closest point(ICP)matching method using Euclidean distance.Mahalanobis
作者
李兆强
张岳
熊福力
苏惠杰
Li Zhaoqiang;Zhang Yue;Xiong Fuli;Su Huijie(School of Information and Control Engineering,Xi’an University of Architecture and Technology,Xi’an 710055,Shaanxi,China;Xi’an Key Laboratory of Smart Industry Perception,Computing and Decisionmaking,Xi’an University of Architecture and Technology,Xi’an 710055,Shaanxi,China)
出处
《光学学报》
EI
CAS
CSCD
北大核心
2024年第20期195-207,共13页
Acta Optica Sinica
基金
陕西省自然科学基础研究计划(2023-JC-YB-582)。
关键词
遥感
无人驾驶
激光雷达同时定位与建图算法
地面分割
回环优化
特征提取
remote sensing
autonomous driving
laser simultaneous localization and mapping algorithm
ground segmentation
loop optimization
feature extraction
