期刊文献+

惯导卫星组合导航辅助的视觉导盲仪定位建图 预览

Localization and Mapping of Visual Travel Aids Assisted by IMU/GPS Integrated Navigation
在线阅读 下载PDF
分享 导出
摘要 视觉即时定位与建图(VSLAM)算法能够为穿戴式视觉导盲仪提供必要的位姿信息和环境信息。针对穿戴式视觉导盲仪对定位与建图精度的要求,提出了一种利用IMU/GPS组合导航辅助VSLAM的算法。在双目ORB-SLAM的跟踪线程中,将匀速运动模型改进为基于IMU/GPS组合导航位姿数据的运动模型,并且构造了综合视觉与组合导航数据的非线性优化,以提高跟踪线程的稳定性、提升定位精度。同时,局部建图线程的地图优化中也融合了组合导航数据,提高局部地图的精度,便于导盲仪检测环境信息,最终得到在大范围场景下稳定运行的定位与建图算法。在实际场景中的试验结果表明,该融合算法抑制了VSLAM的累积误差,同时VSLAM能够有效抑制GPS受环境影响产生的定位偏移误差,最终的定位精度达到0.47 m。该算法提高了VSLAM建图效果,基本满足了视觉导盲仪对定位精度和环境信息的应用需求。 The visual simultaneous localization and mapping(VSLAM)algorithm can provide necessary pose information and environmental information for the wearable visual travel aids(VTA).Aiming at the requirements of positioning and mapping accuracies,a visual simultaneous localization and mapping algorithm aided by IMU/GPS integrated navigation is proposed.In the tracking thread of binocular ORB-SLAM,the uniform motion model is transformed into the motion model based on IMU/GPS integrated navigation data,and the nonlinear optimization that synthesized visual and integrated navigation data is constructed to improve the stability of the tracking thread and improve the positioning accuracy.The integrated navigation data is also integrated into the map optimization of the local mapping thread to improve accuracy of the local map and facilitate the detection of environmental information.A simultaneous location and mapping algorithm that run stably under a large range of environments is finally obtained.The experimental results in the actual environments show that,the fusion algorithm suppresses the accumulative errors and VSLAM can effectively suppress GPS offset error caused by environmental impact.The final positioning accuracy reaches 0.47 meters and the effect of mapping is improved.It can satisfy the visual travel aids requirements of positioning accuracy and environmental information.
作者 魏彤 李新 WEI Tong;LI Xin(School of Instrumentation Science and Opto-Electronics Engineering,Beihang University,Beijing 100191,China)
出处 《自动化仪表》 CAS 2019年第1期45-49,共5页 Process Automation Instrumentation
基金 北京市科技计划基金资助项目(Z151100002115022).
关键词 计算机视觉 定位 建图 非线性优化 多传感器融合 组合导航 可穿戴设备 导盲仪 Computer vision Localization Mapping Nonlinear optimization Multi-sensor fusion Integrated navigation Wearable device Blind-guide instrument
作者简介 魏彤(1972—),男,博士,副教授,硕士生导师,主要从事计算机视觉与控制、机电系统控制工程方向的研究,E-mail:weitong@buaa.edu.cn;通信作者:李新,男,在读硕士研究生,主要从事VSLAM、多传感器融合方向的研究,E-mail:lixin17@buaa.edu.cn
  • 相关文献

参考文献2

二级参考文献5

共引文献87

投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈