安徽工程大学学报2025，Vol.40Issue(2)：21-31,11.
曲线运动下基于关键帧补偿决策的快速SLAM算法
Fast ORB-SLAM Algorithm Based on Keyframe Compensation Decision Under Curved Motion
摘要
Abstract
Aiming at the problems such as low feature tracking and matching time of traditional SLAM algorithm,poor tracking accuracy of optical flow and feature point fusion algorithm,and easy selection of keyframe with poor image quality when the robot moves in curve,an FA-SLAM(Fast SLAM algorithm based on keyframe compensation decision in curvilinear)is proposed.The algorithm fuses LK optical flow and ORB feature points,and adopts the adaptive feature matching strategy with double threshold constraints in optical flow tracking to reduce the computational redundancy,quickly calculates the camera pose,and improves the real-time and tracking accuracy of the system.Meanwhile,in order to avoid that keyframe with poor image quality are easily selected when the robot moves in a curved direction,which leads to insufficient localization accuracy of the system,a keyframe compensation decision to redefine the image quality is proposed,and high-quality images are selected as active keyframe by determining the number of active feature points in the current frame,so as to improve the localization accuracy of the robot.The proposed algorithm is tested on TUM public data set,and the results show that the running time of the proposed algorithm is reduced by 43.1%and the absolute trajectory error is reduced by 31.8%compared with ORB-SLAM2.关键词
同步定位与建图/光流跟踪/双阈值约束/活跃特征点/关键帧补偿Key words
simultaneous localization and mapping/optical tracking/double threshold constraint/active feature points/keyframe compensation分类
计算机与自动化引用本文复制引用
姚满宇,陈孟元,徐韬..曲线运动下基于关键帧补偿决策的快速SLAM算法[J].安徽工程大学学报,2025,40(2):21-31,11.基金项目
安徽省重点研究与开发计划项目(202304a05020073) （202304a05020073）
安徽省高校协同创新项目(GXXT-2021-050) （GXXT-2021-050）
