中文摘要：

针对复杂室外场景,提出一种基于拓扑高程模型的三维环境建模方法.采用自适应可变阈值聚类算法,将映射到二维水平栅格中的激光点云划分为垂直单元和水平单元,可实现三维场景中悬空环境特征的有效表述.在此基础上对垂直单元进行高度离散采样,从而构建与其相对应的拓扑结构,并结合BOW（Bag of words）模型对室外三维环境中的典型景物进行辨识.采用面向拓扑结构和高程图单元的分级匹配策略,实现不同场景间的精确匹配,构建具有全局一致性的拓扑高程地图.利用辨识出来的环境特征和高程地图产生双重环境约束,实现与室外地形相适应的自主路径规划.实验结果和数据分析证明了本文环境建模与路径规划方法的有效性和实用性.

英文摘要：

An approach for 3D environment modeling is proposed to establish the topology-elevation model of outdoor scenes. After projecting laser points into 2D grids,the clustering algorithm classifies the points into vertical units and horizontal units for purpose of describing overhanging structures. The discrete samples of vertical units layered by height are connected to form topological structures,which are utilized to identify typical objects by means of BOW （Bag of words） classifier. A hierarchical strategy,in which topological structures and elevation units are used for scene matching separately,is presented for achieving a globally consistent registration. The constraints extracted from environment features and elevation map are used to guide mobile robot s autonomous path planning in outdoor environment. Experiment results and data analysis show the validity and practicability of the proposed approaches in 3D environment modeling and path planning.