摘要
提出了一种新型拓扑地图;该地图用激光的扇区特征和视觉的比例不变特征来联合表示节点.与传统地图相比,该地图在创建过程中不依赖任何人工路标和机器人的全局定位.机器人通过综合考虑单个节点的相似度和不同节点间的空间关系,利用隐马尔可夫模型来提高节点的识别率.实验表明,该地图不仅易于创建和维护,而且适用于机器人在大规模室内环境下的自主导航.
A novel topological map is proposed in this paper, and its nodes are represented with the visual scale-invariant features and the beam features of the laser range finder. Compared with traditional map presentations, this method doesn't need the robot global localization nor the artificial landmarks to build the map. Both the similarity of single node and the spatial relationship between individual nodes are taken into consideration, and the Hidden Markov Model (HMM) is used to improve the node recognition rate. The navigation experiments demonstrate that the topological map is not only easy to build and maintain, but also suitable for autonomous navigation in large-scale indoor environment without any landmarks.
出处
《机器人》
EI
CSCD
北大核心
2007年第5期433-438,共6页
Robot
基金
国家863计划资助项目(2006AA04Z259)
国家自然科学基金资助项目(60643005)
关键词
地图创建
自主导航
拓扑地图
比例不变特征
隐马尔可夫模型
map building
autonomous navigation
topological map
scale-invariant feature
HMM ( Hidden Markov Model)