Cooperative coverage path planning for visual inspection
Cooperative coverage path planning for visual inspection
复制标题
视觉检测的协同覆盖路径规划
DOI：
10.1016/j.conengprac.2018.03.002
复制
发表时间：
2018-05-01
影响因子：
4.9
通讯作者：
Nikolakopoulos, George
中科院分区：
文献类型：
--
作者：
Mansouri, Sina Sharif;Kanellakis, Christoforos;Nikolakopoulos, George
本文讨论了使用多个无人机（UAV）的复杂3D基础设施的检测问题。所提出的方案的主要新奇源于建立一个理论框架，该框架能够提供一条路径，用于实现基础设施的全覆盖，而无需任何进一步的简化（所考虑的代表点的数量），通过将其由水平平面切片以识别分支并将特定区域分配给每个代理作为整体优化问题的解决方案。此外，在覆盖任务期间收集的图像流使用运动结构、立体SLAM和网格重建算法进行后处理，而所得的3D网格可用于进一步的视觉检查目的。所提出的协作覆盖路径规划（C-CPP）的性能已经在多个室内和现实的室外基础设施检查实验中进行了实验评估，因此它也对无人机的真实的应用做出了重大贡献。
This article addresses the inspection problem of a complex 3D infrastructure using multiple Unmanned Aerial Vehicles (UAVs). The main novelty of the proposed scheme stems from the establishment of a theoretical framework capable of providing a path for accomplishing a full coverage of the infrastructure, without any further simplifications (number of considered representation points), by slicing it by horizontal planes to identify branches and assign specific areas to each agent as a solution to an overall optimization problem. Furthermore, the image streams collected during the coverage task are post-processed using Structure from Motion, stereo SLAM and mesh reconstruction algorithms, while the resulting 3D mesh can be used for further visual inspection purposes. The performance of the proposed Collaborative-Coverage Path Planning (C-CPP) has been experimentally evaluated in multiple indoor and realistic outdoor infrastructure inspection experiments and as such it is also contributing significantly towards real life applications for UAVs.