Scholarlink

<p>In forested environments lacking global navigation satellite system signals, two-dimensional light detection and ranging (2D LiDAR)-based simultaneous localization and mapping (SLAM) suffers degraded performance due to dense vegetation and terrain undulations. To address these challenges, this study proposes a multi-sensor localization and mapping method for forestry mobile platforms based on an enhanced Hector SLAM. An adaptive-parameter density-based spatial clustering of applications with noise preprocesses LiDAR data to suppress vegetation noise and sensor artefacts. Inertial measurement unit (IMU) attitude data corrects LiDAR scan projections, while a pitch-compensated extended kalman filter fuses high-frequency IMU predictions with LiDAR-derived poses, improving terrain adaptability. For localization, point-to-line iterative closest point-based LiDAR odometry replaces wheel odometry. Field experiments in forest settings demonstrate the method’s efficacy: mapping experiments achieve a 4.78 cm average error (47.65% reduction versus original Hector SLAM), with maximum errors below 8.7 cm. Static localization experiments yield a mean lateral error of 9.9 cm, while dynamic localization experiments under three velocity conditions exhibit mean lateral errors below 7.9 cm and mean heading errors less than 36.5°. These results confirm the method's suitability for complex forest environments, providing a cost-effective technical foundation for unmanned forestry operations.</p>

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

Localization and mapping method for forestry mobile platforms based on enhanced Hector SLAM

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月24日

查看全文
  • Liwen Yao,
  • Junhao Wen,
  • Ce Bian,
  • Yuhong Chen,
  • Lijun Xu,
  • Lijian Yao

摘要

In forested environments lacking global navigation satellite system signals, two-dimensional light detection and ranging (2D LiDAR)-based simultaneous localization and mapping (SLAM) suffers degraded performance due to dense vegetation and terrain undulations. To address these challenges, this study proposes a multi-sensor localization and mapping method for forestry mobile platforms based on an enhanced Hector SLAM. An adaptive-parameter density-based spatial clustering of applications with noise preprocesses LiDAR data to suppress vegetation noise and sensor artefacts. Inertial measurement unit (IMU) attitude data corrects LiDAR scan projections, while a pitch-compensated extended kalman filter fuses high-frequency IMU predictions with LiDAR-derived poses, improving terrain adaptability. For localization, point-to-line iterative closest point-based LiDAR odometry replaces wheel odometry. Field experiments in forest settings demonstrate the method’s efficacy: mapping experiments achieve a 4.78 cm average error (47.65% reduction versus original Hector SLAM), with maximum errors below 8.7 cm. Static localization experiments yield a mean lateral error of 9.9 cm, while dynamic localization experiments under three velocity conditions exhibit mean lateral errors below 7.9 cm and mean heading errors less than 36.5°. These results confirm the method's suitability for complex forest environments, providing a cost-effective technical foundation for unmanned forestry operations.