Scholarlink

This study proposed an image feature matching-based method for verifying the speeds of unmanned surface vessels (USVs), validated against Real-Time Kinematic (RTK) positioning measurements. The hardware system was composed of Ground Control Points (GCPs), a DJI MAVIC 3 Pro drone, an iBoat-BS12 USV equipped with an onboard RTK rover module, a DGS680 RTK positioning system, and a computer. Experiments were conducted in a 400 m long and 15 m wide channel, and five cases with speeds of 0.2 ~ 3.0 m/s were tested. Results demonstrated strong consistency between image-derived and RTK-derived speeds. Mean absolute errors (MAE) between the two methods increased linearly with the USV speed (V), fitted as MAE = 0.2469V + 0.2872, while mean relative errors (MRE) mainly fluctuated around 0.5%. Standard deviation (SD) and measurement uncertainty (MU) followed SD = 0.2354V + 0.3426 and MU = 0.5067V + 0.297. The method proved reliability for the verification of low-speed USV movements (≤3 m/s) in static inland waters, offering a viable speed estimation for USV operations without onboard RTK rover modules. Future work will optimize the algorithm for complex hydrodynamic conditions.

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

Speed Verification for Unmanned Surface Vessels Based on Image Feature Matching Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Huai Chen,
  • Junyin Liu,
  • Jianyin Zhou,
  • Lei Zhang,
  • Fanyi Zhang,
  • Jian Jiao,
  • Nairu Wang,
  • Lijun Zhu

摘要

This study proposed an image feature matching-based method for verifying the speeds of unmanned surface vessels (USVs), validated against Real-Time Kinematic (RTK) positioning measurements. The hardware system was composed of Ground Control Points (GCPs), a DJI MAVIC 3 Pro drone, an iBoat-BS12 USV equipped with an onboard RTK rover module, a DGS680 RTK positioning system, and a computer. Experiments were conducted in a 400 m long and 15 m wide channel, and five cases with speeds of 0.2 ~ 3.0 m/s were tested. Results demonstrated strong consistency between image-derived and RTK-derived speeds. Mean absolute errors (MAE) between the two methods increased linearly with the USV speed (V), fitted as MAE = 0.2469V + 0.2872, while mean relative errors (MRE) mainly fluctuated around 0.5%. Standard deviation (SD) and measurement uncertainty (MU) followed SD = 0.2354V + 0.3426 and MU = 0.5067V + 0.297. The method proved reliability for the verification of low-speed USV movements (≤3 m/s) in static inland waters, offering a viable speed estimation for USV operations without onboard RTK rover modules. Future work will optimize the algorithm for complex hydrodynamic conditions.