Scholarlink

To solve the localization problem caused by error accumulation in continuous loop scenarios. This thesis presents a localization approach that boosts the SLAM algorithm’s comprehensive efficacy by integrating multi-sensor loop detection. Firstly, Lidar, IMU, and camera measurement data are fused using an iterative error-state Kalman filtering framework to achieve accurate and consistent state estimation, capitalizing on the unique sensing capabilities of each device. Meanwhile, the key frame selection strategy is optimized according to different system states, and the loop detection is employed to eliminate cumulative errors. Secondly, factor graph optimization is utilized to further enhance the overall system accuracy. Finally, experimental results on public datasets and real indoor scenes demonstrate that the algorithm significantly enhances the accuracy and robustness of pose estimation.

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

A SLAM Loop Closure Detection Method that Fuses Visual and 3D Laser Information

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Qingji Gao,
  • Yunyan Li,
  • Xuanyao Jia

摘要

To solve the localization problem caused by error accumulation in continuous loop scenarios. This thesis presents a localization approach that boosts the SLAM algorithm’s comprehensive efficacy by integrating multi-sensor loop detection. Firstly, Lidar, IMU, and camera measurement data are fused using an iterative error-state Kalman filtering framework to achieve accurate and consistent state estimation, capitalizing on the unique sensing capabilities of each device. Meanwhile, the key frame selection strategy is optimized according to different system states, and the loop detection is employed to eliminate cumulative errors. Secondly, factor graph optimization is utilized to further enhance the overall system accuracy. Finally, experimental results on public datasets and real indoor scenes demonstrate that the algorithm significantly enhances the accuracy and robustness of pose estimation.