摘 要:精确鲁棒的定位系统是保证室内巡检机器人正常工作的重要基础。文章基于机器人操作系统(ROS),针对现有公开视觉算法 ORB-SLAM2 在低性能计算平台上因计算能力不足导致的特征跟踪丢失问题,提出一种将 ORB-SLAM2 与惯性导航系统(Inertial Navigation System, INS)解算误差进行卡尔曼滤波融合的方法。经公开数据集验证表明,该方法能够完整地估计出视觉失效时丢失的位姿信息,与 ORB-SLAM2 相比,定位系统的精度与鲁棒性有效提高。
关键词:ROS;定位;卡尔曼滤波;ORB-SLAM2
DOI:10.19850/j.cnki.2096-4706.2021.13.036
基金项目:国家自然科学基金(51879046); 黑龙江省自然科学基金 (YQ2019F001)
中图分类号:TP242 文献标识码:A 文章编号:2096-4706(2021)13-0139-06
Visual Inertial Fusion Positioning Method Based on ROS Indoor Inspection Robot
SUN Xijun1 , WANG Qiuying2,3 , WANG Shuigen4 , WU Yingwei 1
(1.College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China; 2.College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, China; 3.Key Laboratory of Underwater Acoustic Technology, Harbin Engineering University, Harbin 150001, China; 4.Yantai Iray Technology Co., Ltd., Yantai 264006, China)
Abstract: Accurate and robust positioning system is an important basis to ensure the normal operation of indoor inspection robot. Based on the robot operating system (ROS), aiming at the loss of feature tracking caused by the insufficient computing power of the existing public vision algorithm ORB-SLAM2 on the low-performance computing platform, this paper proposes a method for Kalman filter fusion in the solution errors of ORB-SLAM2 and inertial navigation system (INS). The verification of public data sets shows that this method can completely estimate the pose information lost when visual failure. Compared with ORB-SLAM2, the accuracy and robustness of the positioning system are effectively improved.
Keywords: ROS; positioning; Kalman filter; ORB-SLAM2
参考文献:
[1] 杨观赐,王霄远,蒋亚汶,等 . 视觉与惯性传感器融合的 SLAM 技术综述 [J]. 贵州大学学报(自然科学版),2020,37(6): 1-12.
[2] ENGEL J,KOLTUN V,CREMERS D. Direct Sparse Odometry [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,40(3):611-625.
[3] MUR-ARTAL R,TARDÓS J D. ORB-SLAM2:AnOpenSource SLAM System for Monocular,Stereo,and RGB-D Cameras [J]. IEEE Transactions on Robotics,2017,33(5):1255-1262.
[4] 黄剑雄,刘小雄,章卫国,等 . 基于视觉 / 惯导的无人机组合导航算法研究 [J]. 计算机测量与控制,2021,29(2):137-143+149.
[5] 杜小菁,陈洪,王欣,等 . 基于 SINS 解算的行人导航技术发展综述 [J]. 战术导弹技术,2020(1):113-120.
[6] 孟春见,李开龙,张梦得 . 捷联惯性基组合导航滤波算法比较研究 [J]. 电光与控制,2020,27(2):18-21.
作者简介:孙希君(1997—),女,汉族,山东武城人,硕士 研究生在读,研究方向:传感器融合定位、组合导航。