Scholarlink

<p>To address the challenges of visual perception in quadrotor unmanned aerial vehicle (UAV) autonomous landing caused by limitations in camera viewing angles and visual range, this paper proposes a multi-stage vision-based localization method for UAV autonomous landing. The method achieves coarse localization of the UAV through ground feature extraction and matching, followed by precise localization using cooperative markers, enabling accurate landing. In the coarse localization stage, the DISK and LightGlue algorithms are employed to extract and match ground feature points, while the UAV’s position is preliminarily estimated using the known distance between the camera and the ground. In the precise localization stage, the AprilTag algorithm is utilized to detect cooperative markers on the landing platform, further enhancing landing accuracy. Experimental results demonstrate that the proposed method exhibits strong robustness and high localization accuracy in complex environments, effectively achieving vision-assisted precise landing for UAVs.</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

Multi-Stage Visual Guidance for UAV Autonomous Landing

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • Hongyu Wang,
  • Guangyu Dong,
  • Zheng Dang,
  • Cheng Cheng,
  • Fei Chen

摘要

To address the challenges of visual perception in quadrotor unmanned aerial vehicle (UAV) autonomous landing caused by limitations in camera viewing angles and visual range, this paper proposes a multi-stage vision-based localization method for UAV autonomous landing. The method achieves coarse localization of the UAV through ground feature extraction and matching, followed by precise localization using cooperative markers, enabling accurate landing. In the coarse localization stage, the DISK and LightGlue algorithms are employed to extract and match ground feature points, while the UAV’s position is preliminarily estimated using the known distance between the camera and the ground. In the precise localization stage, the AprilTag algorithm is utilized to detect cooperative markers on the landing platform, further enhancing landing accuracy. Experimental results demonstrate that the proposed method exhibits strong robustness and high localization accuracy in complex environments, effectively achieving vision-assisted precise landing for UAVs.