Abstract: In recent years,due to the development of multi-robot optical technology,the coverage inspection of large 3D structures like automobile and other large components comes into reality.Besides,it also lays the foundation for the quality evaluation of whole surfaces of products and process optimization.Because of the complexity of solving multiple NP-Hard problems such as full-coverage viewpoint sampling of large structural parts,multi-robot task assignment and path planning,the application of multi-robot coverage path planning still relies on experience.Therefore,a new method of multi-robot path planning for the coverage inspection of large-scale 3D structures was proposed.The initial redundant viewpoints set were generated according to the distribution of to-be-inspect geometric features,and the greedy algorithm was employed for the optimal sampling of full coverage.Afterwards,an integrated optimization model for task assignment and viewpoint sequence planning was constructed.The body-in-white case was used to verify the effectiveness of the proposed method.The Result showed that the average inspection cycle time of the multi-robot inspection cell was reduced by 31.56%,and the run-time consistency of multi-robot was reduced by 13.13%.

Key words: 3D structure, quality inspection, multi-robot system, coverage path planning

摘要： 多机器人覆盖检测技术在汽车、航空航天部件等大型三维结构检测中的推广应用,为产品的全型面质量评价与工艺优化提供了依据。然而,受到大型结构件全覆盖视点采样、多机任务分配、路径规划等多重NP-hard问题求解的复杂性限制,多机器人覆盖路径规划的工程应用仍主要依赖经验,缺乏有效的理论依据。为此,针对大型三维结构的覆盖检测问题,提出一种多机器人检测路径规划的新方法。首先,根据待测几何特征分布生成初始冗余视点集合,采用贪心算法实现全覆盖视点的优化采样;在此基础上,构建了面向多机器人视点任务分配与视点次序规划的集成优化模型,并通过将多机器人可达性、视点次序、距离矩阵等多源信息编码,提出基于改进遗传算法的覆盖路径规划问题的求解新方法。最后,通过白车身结构的仿真案例与对比分析,验证所提方法的有效性。结果表明,基于所提方法的工位内检测周期平均下降31.56%,多机检测一致性提升13.13%。

关键词: 三维结构, 质量检测, 多机器人系统, 覆盖路径规划