Research on Visual Guided 3D Tracking Control for UUV Docking
-
摘要: 【目的】自主对接是水下无人航行器(UUV)协同作业的关键，但受复杂环境和对象特性影响，精准引导与对接难度很大。为了提高水下对接的准确性和鲁棒性，本文设计了一种基于视觉引导的对接方案，并针对视觉解算和三维跟踪控制技术开展研究。【方法】首先，结合任务和对象特性分析，设计了基于视觉引导的总体对接方案。其次，设计YOLOv5神经网络完成了水下对接站的目标检测，并基于EPnP算法实现了对接站与UUV间相对位姿关系的在线测量。接着，结合视觉解算结果，基于三维LOS制导、RBF神经网络、终端滑模控制和李雅普诺夫理论，设计了一种高效的三维鲁棒跟踪控制器。最后，通过数字仿真和水池试验验证本文设计方案的有效性。【结果】在水池试验中，视觉引导控制算法能有效完成水下对接站的在线检测与相对定位，且实现UUV精准水下对接。【结论】研究表明，本文提出的视觉引导三维跟踪控制方案合理、高效，可为UUV水下对接奠定基础，且该技术具有广阔的应用前景。Abstract: [Objectives] Autonomous docking is the key to the cooperative operation of underwater unmanned vehicles (UUV). However, due to the complex environment and object characteristics, precise guidance and docking are very difficult. In order to improve the accuracy and robustness of underwater docking, a vision guided docking scheme is designed in this paper, where vision processing and 3D tracking control are studied. [Methods] Firstly, the overall visual guided docking scheme is designed combining with the analysis of task and object characteristics. Secondly, YOLOv5 neural network is designed to complete the target detection of underwater docking station, and the online measurement of the relative position and attitude relationship between docking station and UUV is realized based on EPnP (Efficient perspective-n-point) algorithm. Then, combined with
-
Key words:
- UUV /
- Underwater Docking /
- Visual Guidance /
- 3D Tracking /
- Pool test
-
计量
- 文章访问数: 116
- 被引次数: 0
下载:
