基于图优化的Kinect三维视觉里程计设计
|
摘要
针对Kinect相机在未知场景中的路径估计问题,提出了一种基于图优化的视觉里程计算法。通过深度图像的光流匹配筛选出关键帧,并得到关键帧的初始位姿估计;将关键帧的初始位姿估计作为顶点,位姿之间的变换作为边构成一个连通图模型,并通过回环检测在图上增加回环;在连通图模型上利用非线性最小二乘对初始位姿优化,从而得到视觉里程计。实验结果表明:提出的方法在满足实时性的基础上,有效减少了误差,这在以视觉里程计为基础的应用中具有很重要的作用。
Aiming at the problem of path estimation of Kinect camera in unknown scene,a visual odometry based on graph optimization is proposed. Key frame is screened out by optical flow matching of depth images,and the initial pose estimation of the key frame is obtained. Initial pose estimation of the key frame is taken as the vertex and the transformation between the poses as side to form a connected graph model. And the loopback is added to the graph by loopback detection. Non-linear least squares is used to optimize the initial pose in the graph model,and the visual odometer is obtained. The experimental results show that the method can reduce the error effectively on the basis of meeting real-time performance,which plays an important role in the application based on visual odometer.
Aiming at the problem of path estimation of Kinect camera in unknown scene,a visual odometry based on graph optimization is proposed. Key frame is screened out by optical flow matching of depth images,and the initial pose estimation of the key frame is obtained. Initial pose estimation of the key frame is taken as the vertex and the transformation between the poses as side to form a connected graph model. And the loopback is added to the graph by loopback detection. Non-linear least squares is used to optimize the initial pose in the graph model,and the visual odometer is obtained. The experimental results show that the method can reduce the error effectively on the basis of meeting real-time performance,which plays an important role in the application based on visual odometer.
引文
[1] Leonard J J,Durrantwhyte H F. Mobile robot localization bytracking geometric beacons[J]. IEEE Transactions on Robotics&Automation,1991,7(3):376-382.
[2]武二永,项志宇,沈敏一,等.大规模环境下基于激光雷达的机器人SLAM算法[J].浙江大学学报:工学版,2007,41(12):1982-1986.
[3] Nister D,Naroditsky O,Bergen J. Visual odometry[C]∥Procee-dings of the 2004 IEEE Computer Society Conference on Com-puter Vision and Pattern Recognition,CVPR 2004,2004:I-652—I—659.
[4] Bailey T,Nieto J,Guivant J,et al. Consistency of the EKF-SLAMalgorithm[C]∥2006 IEEE/RSJ International Conference onIntelligent Robots and Systems,IEEE,2006:3562-3568.
[5] Montemerlo M,Thrun S,Koller D,et al. Fast SLAM:A factoredsolution to the simultaneous localization and mappingproblem[C]∥The Eighteenth National Conference on ArtificialIntelligence,American Association for Artificial Intelligence,2002:2004.
[6]张栋,焦嵩鸣,刘延泉.互补滤波和卡尔曼滤波的融合姿态解算方法[J].传感器与微系统,2017,36(3):62-65.
[7] Henry P,Krainin M,Herbst E,et al. Experimental robotics[M].Berlin Heidelberg:Springer,2014:647-663.
[8] Martinez Henry P,Krainin M,Herbst E,et al. RGB-D mapping:Using Kinect-style depth cameras for dense 3D modeling of in-door environments[J]. International Journal of RoboticsResearch,2012,31(5):647-663.
[9] Jordan J,Zell A. Ground plane based visual odometry for RGBD-cameras using orthogonal projection[J]. IFAC-Papers Online,2016,49(15):108-113.
[10]杨杰.融合IMU与Kinect的机器人定位算法研究[D].武汉:武汉科技大学,2014.
[11]王亚龙,张奇志,周亚丽.基于Kinect的三维视觉里程计的设计[J].计算机应用,2014,34(8):2371-2374,2379.
[12] Nikolas Engel Hard,Felix Endres,JürgenHess,et al. Real-time3D visual SLAM witha hand-held RGB-D camera[C]∥Proc ofthe RGB-D Workshop on 3D Perception in Robotics at the Euro-pean Robotics Forum,Vsters,Sweden,2016.
[13] Engel J,Schops T,Cremers D. LSD-SLAM:Large-scale directmonocular SLAM[C]∥European Conf on Computer Vision,EC-CV 2014,2014:834-849.
[14] Mur-Artal R,Montiel J M M,Tardós J D. ORB-SLAM:A versatileand accurate monocular SLAM system[J]. IEEE Transactions onRobotics,2015,31(5):1147-1163.
[15]徐军,孟月霞,王天伦,等.基于Kinect的仿人机器人控制系统[J].传感器与微系统,2017,36(9):97-100.
[16]王依人,邓国庆,刘勇,等.基于激光雷达传感器的RBPF-SLAM系统优化设计[J].传感器与微系统,2017,36(9):77-80.
[2]武二永,项志宇,沈敏一,等.大规模环境下基于激光雷达的机器人SLAM算法[J].浙江大学学报:工学版,2007,41(12):1982-1986.
[3] Nister D,Naroditsky O,Bergen J. Visual odometry[C]∥Procee-dings of the 2004 IEEE Computer Society Conference on Com-puter Vision and Pattern Recognition,CVPR 2004,2004:I-652—I—659.
[4] Bailey T,Nieto J,Guivant J,et al. Consistency of the EKF-SLAMalgorithm[C]∥2006 IEEE/RSJ International Conference onIntelligent Robots and Systems,IEEE,2006:3562-3568.
[5] Montemerlo M,Thrun S,Koller D,et al. Fast SLAM:A factoredsolution to the simultaneous localization and mappingproblem[C]∥The Eighteenth National Conference on ArtificialIntelligence,American Association for Artificial Intelligence,2002:2004.
[6]张栋,焦嵩鸣,刘延泉.互补滤波和卡尔曼滤波的融合姿态解算方法[J].传感器与微系统,2017,36(3):62-65.
[7] Henry P,Krainin M,Herbst E,et al. Experimental robotics[M].Berlin Heidelberg:Springer,2014:647-663.
[8] Martinez Henry P,Krainin M,Herbst E,et al. RGB-D mapping:Using Kinect-style depth cameras for dense 3D modeling of in-door environments[J]. International Journal of RoboticsResearch,2012,31(5):647-663.
[9] Jordan J,Zell A. Ground plane based visual odometry for RGBD-cameras using orthogonal projection[J]. IFAC-Papers Online,2016,49(15):108-113.
[10]杨杰.融合IMU与Kinect的机器人定位算法研究[D].武汉:武汉科技大学,2014.
[11]王亚龙,张奇志,周亚丽.基于Kinect的三维视觉里程计的设计[J].计算机应用,2014,34(8):2371-2374,2379.
[12] Nikolas Engel Hard,Felix Endres,JürgenHess,et al. Real-time3D visual SLAM witha hand-held RGB-D camera[C]∥Proc ofthe RGB-D Workshop on 3D Perception in Robotics at the Euro-pean Robotics Forum,Vsters,Sweden,2016.
[13] Engel J,Schops T,Cremers D. LSD-SLAM:Large-scale directmonocular SLAM[C]∥European Conf on Computer Vision,EC-CV 2014,2014:834-849.
[14] Mur-Artal R,Montiel J M M,Tardós J D. ORB-SLAM:A versatileand accurate monocular SLAM system[J]. IEEE Transactions onRobotics,2015,31(5):1147-1163.
[15]徐军,孟月霞,王天伦,等.基于Kinect的仿人机器人控制系统[J].传感器与微系统,2017,36(9):97-100.
[16]王依人,邓国庆,刘勇,等.基于激光雷达传感器的RBPF-SLAM系统优化设计[J].传感器与微系统,2017,36(9):77-80.