摘要
构建了一个融合特征法与直接法的快速、鲁棒同时定位与地图构建(SLAM)系统CFD-SLAM,该系统能够同时应用于单目、双目和RGB-D相机。SLAM系统主要由3部分组成追踪、局部建图和闭环检测。追踪部分融合特征法与直接法,并对关键帧和非关键帧分别采用特征法和直接法进行追踪,以提高系统的实时性和在特征缺失环境下的鲁棒性。特征法提取ORB特征和计算BRIEF描述子,并通过最小化特征点的重投影误差来获得关键帧的位姿估计。直接法通过最小化光度误差来获得非关键帧的位姿估计。对于RGB-D相机,逆深度误差被加入到特征法和直接法的优化目标函数中。局部建图部分负责管理局部关键帧和地图点,并通过光束平差法(BA)来优化局部关键帧位姿和局部地图点位置。对于RGB-D相机,该SLAM系统能够构建八叉树地图,用于机器人导航等高级任务。闭环检测部分通过检测闭环关键帧和执行位姿图优化来提高SLAM系统的全局一致性。本文通过在开源数据集上与典型开源SLAM系统进行对比实验,证明了本文的SLAM系统在保证定位精度的同时,具有较好的实时性和鲁棒性。
This work proposes a fast and robust simultaneous localization and mapping(SLAM)system combining feature-based method and direct method(CFD-SLAM),which can be used for monocular,stereo and RGB-D cameras.The SLAM system consists of 3 parts:tracking,local mapping and loop closure.The tracking part combines feature-based method and direct method,feature-based method and direct method are utilized for keyframes and non-keyframes tracking respectively to improve real-time performance and robustness in low-texture environments.Feature-based method extracts ORB(oriented FAST and rotated BRIEF)features and computes BRIEF(binary robust independent elementary features)descriptors,and obtains the pose estimation of keyframes by minimizing re-projection error of feature points.Direct method obtains the pose estimation of non-keyframes by minimizing photometric error.For RGB-D cameras,inverse depth error is added into optimization cost function of both feature-based method and direct method.The local mapping part is in charge of managing local keyframes and map points,and optimizing both poses of local keyframes and position of local map points by using bundle adjustment(BA).To perform high level tasks such as navigation,the Octomap of the environment can be constructed for RGB-D cameras.The loop closure part improves the global consistency of the SLAM system by detecting loop keyframes and executing pose graph optimization.Finally,experiments on public datasets against other state-of-the-art SLAM systems demonstrate that the system is faster and more robust than the state-of-the-art SLAM system without precision reduction.
作者
王化友
代波
何玉庆
Wang Huayou;Dai Bo;He Yuqing(State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang 110016;University of Chinese Academy of Sciences,Beijing 100049;Shenyang Institute of Automation(Guangzhou),Chinese Academy of Sciences,Guangzhou 511458)
出处
《高技术通讯》
EI
CAS
北大核心
2019年第12期1224-1238,共15页
Chinese High Technology Letters
基金
国家自然科学基金(61433016,61503369)
广东省科技计划(2017B010116002)资助项目
