Scholarlink

Visual simultaneous localization and mapping (SLAM) systems still have limitation in application due to the range-limited depth estimation of visual cameras in large-scale motion-distance scenes. The accurate depth estimation capability of LiDAR in 3D scenes makes it adventageous to compensate the defects of camera. This study designs a novel strategy to correlate the visual ORB feature points with the corresponding LiDAR depth information so as to generate the fused feature points. Based on this fusion strategy, the Visual/LiDAR/Inertial Measurement Unit (IMU)-based ORB-SLAM3 is further developed, which extends the applicability of the original ORB-SLAM3. Finally, the proposed SLAM system is evaluated based on open source datasets in terms of the stability of the fused feature points and the localization accuracy of the extended system using the fused feature points.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Feature Point Fusion Strategy Based on Visual and LIDAR Information

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shijie Wu,
  • Haoyu Qi,
  • Yuhang Zhang,
  • Zhen Li,
  • Wenjie Chen

摘要

Visual simultaneous localization and mapping (SLAM) systems still have limitation in application due to the range-limited depth estimation of visual cameras in large-scale motion-distance scenes. The accurate depth estimation capability of LiDAR in 3D scenes makes it adventageous to compensate the defects of camera. This study designs a novel strategy to correlate the visual ORB feature points with the corresponding LiDAR depth information so as to generate the fused feature points. Based on this fusion strategy, the Visual/LiDAR/Inertial Measurement Unit (IMU)-based ORB-SLAM3 is further developed, which extends the applicability of the original ORB-SLAM3. Finally, the proposed SLAM system is evaluated based on open source datasets in terms of the stability of the fused feature points and the localization accuracy of the extended system using the fused feature points.