微型机与应用
微型機與應用
미형궤여응용
Microcomputer & its Applications
2015年
16期
15-17,20
,共4页
SLAM%单目视觉%FAST%压缩扩展卡尔曼滤波
SLAM%單目視覺%FAST%壓縮擴展卡爾曼濾波
SLAM%단목시각%FAST%압축확전잡이만려파
Simultaneous Localization and Mapping (SLAM)%monocular vision%Features from Accelerated Segment Test (FAST)%Compressed Extend Kalman Filtering (CEKF)
针对目前 SLAM 算法实时性和鲁棒性的问题，提出了一种改进的实时单目视觉 SLAM算法。该算法采用一个摄像头作为外部传感器来提取机器人行进过程中周围环境的特征信息，用实时性良好的 FAST 提取环境特征点，结合逆深度参数化进行特征点非延时初始化，用压缩扩展卡尔曼滤波更新地图。实验研究表明，该方法提高了算法的鲁棒性和实时性。
針對目前 SLAM 算法實時性和魯棒性的問題，提齣瞭一種改進的實時單目視覺 SLAM算法。該算法採用一箇攝像頭作為外部傳感器來提取機器人行進過程中週圍環境的特徵信息，用實時性良好的 FAST 提取環境特徵點，結閤逆深度參數化進行特徵點非延時初始化，用壓縮擴展卡爾曼濾波更新地圖。實驗研究錶明，該方法提高瞭算法的魯棒性和實時性。
침대목전 SLAM 산법실시성화로봉성적문제，제출료일충개진적실시단목시각 SLAM산법。해산법채용일개섭상두작위외부전감기래제취궤기인행진과정중주위배경적특정신식，용실시성량호적 FAST 제취배경특정점，결합역심도삼수화진행특정점비연시초시화，용압축확전잡이만려파경신지도。실험연구표명，해방법제고료산법적로봉성화실시성。
Aiming at the problem of robust SLAM algorithm, an improved real-time monocular visual SLAM algorithm is proposed. Taking a camera as its external sensor, this algorithm can obtain the feature information of the surrounding. It obtains the feature points of environment by using FAST which has good real-time capability. With the combination of inverse depth parametrization, the feature points begin to initialize without delay. The map is updated by the CEKF. The experimental study shows that this method improves the robustness and real-time of the algorithm.
