作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2025, Vol. 51 ›› Issue (12): 357-367. doi: 10.19678/j.issn.1000-3428.0069718

• 开发研究与工程应用 • 上一篇    下一篇

面向屋顶巡检任务的无人机覆盖路径规划算法

李海旺1, 王澳强1, 张世龙1, 郭彩玲2,*(), 李柏林3   

  1. 1. 西南交通大学唐山研究院, 河北 唐山 063000
    2. 唐山学院河北省智能装备数字化设计及过程仿真重点实验室, 河北 唐山 063000
    3. 西南交通大学机械工程学院, 四川 成都 610000
  • 收稿日期:2024-04-09 修回日期:2024-06-10 出版日期:2025-12-15 发布日期:2025-12-16
  • 通讯作者: 郭彩玲
  • 基金资助:
    河北省高层次人才项目(A202001095); 河北省科技厅重点研发计划(20327407D)

Coverage Path Planning Algorithm for Unmanned Aerial Vehicles in Roof Inspection Tasks

LI Haiwang1, WANG Aoqiang1, ZHANG Shilong1, GUO Cailing2,*(), LI Bailin3   

  1. 1. Tangshan Institute, Southwest Jiaotong University, Tangshan 063000, Hebei, China
    2. Key Laboratory of Intelligent Equipment Digital Design and Process Simulation, Tangshan University, Tangshan 063000, Hebei, China
    3. School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610000, Sichuan, China
  • Received:2024-04-09 Revised:2024-06-10 Online:2025-12-15 Published:2025-12-16
  • Contact: GUO Cailing

摘要:

针对现有三维覆盖路径规划(CPP)算法存在求解效率低的问题, 提出一种基于视点采样的三维CPP算法。首先, 构建一种基于栅格降采样的视图规划模型, 将三维建筑屋顶分解至二维空间进行栅格采样提取视点, 并根据平面法向量特征信息计算视线方向, 通过组合视点和视线参数得到建筑屋顶视图数据, 有效消除视图数据的无序现象; 然后, 设计一种基于贪婪搜索的路径规划方案, 利用往复式和螺旋式覆盖路径的特点对视点访问顺序进行贪婪搜索, 并以路径长度、累计转角和高度变化作为评价指标进行优化求解, 生成建筑屋顶的最优覆盖路径, 显著提高路径搜索效率; 最后, 结合八叉树地图模型和覆盖路径信息进行冲突路径的碰撞检测, 并利用视点插值法优化冲突路径。选取平顶型、人字型、四坡型及其复合型建筑屋顶共8组数据进行CPP实验, 结果表明, 所提算法可实现理想的建筑屋顶CPP效果, 在保证覆盖路径综合评价指标最优的同时, 极大地提高了路径规划的求解效率。

关键词: 机载LiDAR, 屋顶巡检, 视图规划, 路径规划, 贪婪搜索

Abstract:

To address the problem of low solving efficiency in existing 3D Coverage Path Planning (CPP) algorithms, a novel 3D CPP algorithm based on viewpoint sampling is proposed. First, a view planning model based on grid downsampling is constructed. This model decomposes a 3D building roof into a 2D space for grid sampling to extract viewpoints. It determines the line of sight direction based on plane normal information. By combining viewpoint and line of sight parameters, building roof view data are obtained, effectively eliminating the disorder in the view data. Subsequently, a path planning scheme based on greedy search is designed. This scheme uses the characteristics of reciprocating and spiral coverage paths to perform a greedy search in the order of viewpoint visits and optimizes the solution using path length, cumulative turning angle, and height variation as evaluation metrics, thus generating the optimal coverage path for the building roof and significantly improving the path search efficiency. Finally, the octree map model and coverage path information are used for collision detection of conflicting paths and the viewpoint interpolation method is used to optimize the conflicting paths. CPP experiments are conducted using eight sets of data for flat, gable, hipped, and composite building roofs. The results show that the proposed algorithm can achieve ideal building roof CPP, significantly improving the solving efficiency of path planning while ensuring optimal comprehensive evaluation metrics for the coverage path.

Key words: airborne LiDAR, roof inspection, view planning, path planning, greedy search