基于海洋遥感图像的无人艇路径覆盖规划算法

Coverage path planning algorithm of unmanned surface vehicle based on ocean remote sensing images

遥感技术和无人艇的结合在海洋覆盖应用中具有巨大的潜力,提出了一种基于海洋遥感图像的无人艇路径覆盖方法。首先,为了建立精确的地图模型,提出了一种基于改进YOLO V3的旋转目标检测算法,在YOLO V3的基础上,细化障碍物的轴向、长度、宽度和坐标信息,在不增加计算量的情况下提高复杂场景下障碍物检测的召回率。然后,为了获得高效的覆盖路径,提出了一种基于旋转光束和贪心算法的路径覆盖算法。该算法将完整路径分为直行路径和转弯路径,分别基于长度与避障目标优化覆盖路径。仿真结果表明,较基于栅格地图的神经元激励算法,所提出算法在长度上平均减少了9.3%,并且在2种极端海洋环境中实现路径覆盖率为100%,重复率小于2.1%。

The combination of remote sensing technology and unmanned surface vehicle has great potential in ocean coverage applications.A coverage path planning(CPP)algorithm for unmanned surface vehicle based on ocean remote sensing images is proposed.Firstly,to establish an accurate map model,a rotating target detection algorithm based on improved YOLO V3 is proposed.Based on YOLO V3,the axis,length,width,and coordinate information of obstacles are refined to improve the recall rate of target detection in complex scenes without increasing the amount of calculation.Then,to obtain effective coverage path,a CPP algorithm based on rotating beams and greedy algorithm is proposed.The algorithm divides the complete path into straight paths and turning paths,and optimizes the coverage path based on the length and obstacle avoidance objectives respectively.Simulation results show that,compared with the neuronal excitation algorithm based on grid map,the length of this algorithm is reduced by 9.3%,and the coverage rate is 100%and the repetition rate is less than 2.1%in two extreme ocean environments.