基于二维目标的增强检测与舷角约束导引方案

Enhanced detection and guidance with side angle constraints based on 2D target

针对无人水面船不确定水域的特定目标搜索与可控角度导引问题,探索了基于YOLO的目标检测方式以及基于舷角约束的导引控制方法。在分析水面海雾等不确定条件基础上,设计相关滤波器并采用改进式YOLOv5算法对目标特征提取,并结合相机深度信息进行坐标转换实现对水面目标的检测定位;在导引规划中,改进传统比例导引为滑模变结构导引控制,实现高机动性目标舷角约束制导。控制系统设计上,采用TCP/IP结合ROS机制,打造了针对视觉信息处理的舷角约束末端制导系统。最终借助Autolabor无人系统平台完成了二维模拟实验验证,目标导引误差结果为1.54%,证明了该方法的有效性和可行性。

Aiming at the specific target search and controllable angle guidance of unmanned surface ships in uncertain waters,this paper explores the target detection method based on YOLO and the guidance control method based on side angle constraint.On the basis of analyzing the uncertain conditions such as sea fog,the paper designs the correlation filter with improved YOLOv5 algorithm to extract the target features,combines depth information to realize the detection and positioning of water surface targets,sliding mode variable structure guidance control to achieve the high maneuverability target side angle constraint guidance.TCP/IP combined with ROS is used to create a corner constrained terminal guidance system for information processing.Two-dimensional simulation experiment is completed with Autolabor platform,and the result of target guidance error is 1.54%,which proves the effectiveness of the method proposed.