摘要
目前,视觉机器人的同时定位与地图构建(Simultaneous Localization and Mapping,SLAM)算法的建图部分主要采用了三维八叉树地图,虽然其地图存储容量较大,但是地图范围却无法实时扩大,且室内场景中常见的动态事物也因为忽略了大噪声点而难以进行处理.为此提出一种新的基于生长型四叉树结构的实时网格二维地图构建方法,将三维体素地图降维为二维网格地图,增加了对动态特征点轨迹的预测,在无损三维空间信息的情况下丰富了导航地图携带的环境信息.真实室内场景的实验表明:该算法能够在地图中较为精确地显示障碍物的位置信息,显著地降低了地图存储空间,提高了建图速度.
The mapping part in the simultaneous localization and mapping(SLAM)algorithm for visual robot positioning and environmental map construction mainly uses the three-dimensional octree map with a large map storage capacity.As a result,the range of mapping cannot be extended adaptively in real time.In the positioning part,it is difficult to deal with the common dynamic things in the indoor scene because of neglecting the large noise points.To address these problems,a novel real-time grid two-dimensional map construction method based on the growth quadtree structure is proposed in this paper to reduce the dimension of the three-dimensional voxel map to the two-dimensional grid map.The prediction of the dynamic feature point trajectory is improved by the proposed method,and the environmental information carried by navigation map is enriched without loss of the three-dimensional spatial information.An experiment in the real indoor scene shows that the proposed algorithm can display the position information of obstacles accurately in the map,greatly reduce the map storage space and improve the drawing speed.
作者
姜晗
贺付亮
王世元
JIANG Han;HE Fu-liang;WANG Shi-yuan(School of Electronic Information Engineering,Southwest University,Chongqing 400715,China;Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing,Chongqing 400715,China)
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第6期128-139,共12页
Journal of Southwest University(Natural Science Edition)
基金
国家自然科学基金项目(61671389)
中央高校基本科研业务费专项资金项目(XDJK2019B011,SWU118060).
关键词
同时定位与建图
场景重建
自主导航
网格地图
四叉树
simultaneous localization and mapping(SLAM)
scene re-construction
autonomous navigation
grid map
quadtree
