Abstract: For the quality inspection of auto body's sheet metal parts,a structured light view planning method was proposed with control on the measurement uncertainty.To achieve coverage inspection of the parts,the proposed method mainly included three stages.In first stage,combined with the measurement uncertainty of each primitive on the surfaces,the admissible viewpoint sampling space was constructed considering the constraints such as incident angle and Depth of Field(DOF).In second stage,the random sampling method was used to obtain the initial viewpoints of the given primitive.Afterwards,with the visibility matrix for the determined viewpoint,the optimal viewpoint set that satisfied the measurement uncertainty of each primitive was generated.In third stage,an improved greedy algorithm was proposed to realize the covering path planning of the free-form surfaces.The feasibility and effectiveness of the method were verified through the simulation and engineering cases.Results showed that the proposed method could effectively improve the inspection accuracy of structured light and reduce the number of viewpoints.

Key words: auto body, 3D optical inspection, measurement uncertainty, viewpoint sampling, coverage path planning

摘要： 针对车身钣金件的质量检测,提出一种考虑曲面上待测特征测量不确定度的结构光视点采样与覆盖路径规划方法。为实现零部件上特征的全覆盖测量,所提方法主要分为3个阶段。首先,结合零部件上各待测特征的测量不确定度要求,生成了考虑入射角、景深等约束的视点可行采样空间;然后,采用随机采样法进行视点的初始采样,并结合视点下的可视化矩阵,生成满足各图元测量不确定度要求的优选视点集合;在此基础上,提出了基于改进贪心算法的集合覆盖问题求解与路径规划方法,实现曲面上待测特征的覆盖检测。最后,通过仿真分析与工程实例验证了该方法的可行性和有效性。结果表明,所提方法能有效提升结构光检测精度、降低视点数目。

关键词: 车身曲面, 3D光学检测, 测量不确定度, 视点采样, 覆盖路径规划