RGB-D Image-based Pose Estimation with Monte Carlo Localization | |
Li, Ming ; Qin, Hao ; Huang, May ; Cao, Jian ; Zhang, Xing | |
2017 | |
关键词 | pose estimation coordinate transformation least squares method monte carlo localization |
英文摘要 | In this paper, a robot vision-based method for localization is presented. We integrate a novel RGB-D imagebased pose estimating technique with Monte Carlo localization to make the mobile robot find its position through image matching and pose correction. A database containing color image with corresponding depth image is pre-generated. The position and orientation information within a map is also included in the database. Color image matching is resolved using a keypoint-matching technique based on SIFT features. Displacement calculation is performed using matching results and depth information. An optimization technique is applied on the displacement calculation to find a more accurate pose. Feeding the pose to Monte Carlo localization, the further precise result is obtained. The method demonstrates improved estimation comparing to pure image matching approach. The experimental results are presented on this paper to illustrate the effectiveness and improvement of this method.; International Technological University (ITU); CPCI-S(ISTP); 109-114 |
语种 | 英语 |
出处 | 3rd IEEE International Conference on Control, Automation and Robotics (ICCAR) |
内容类型 | 其他 |
源URL | [http://ir.pku.edu.cn/handle/20.500.11897/469880] |
专题 | 软件与微电子学院 |
推荐引用方式 GB/T 7714 | Li, Ming,Qin, Hao,Huang, May,et al. RGB-D Image-based Pose Estimation with Monte Carlo Localization. 2017-01-01. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论