Localization of weakly rigid work-piece is converted to rigid registration of two-point clouds by using flexible tooling system.Iterative closest point (ICP)is widely used in rigid registration.But its performance decreases apparently when the rela-tive offset of two-point clouds is large,because of possible local minimum.Good initial position can be guaranteed through manual operation,which relies on operator’s experience.And it is much harder to precisely localize a semi-finished work-piece due to the error and lack of information.In this paper,a edge detection based pre-process algorithm is proposed.By using a small amount of edge points,a transform direction is calculated to lead the work-piece to the better position,which guarantees the ICP prerequi-site.The results show that the combined method could reduce the negative impact of missing parts to ICP,and gain better sensor array coverage.%利用弹性工件刚性化加工技术，可使弱刚度曲面工件的定位问题转化为点云的刚体配准问题。迭代最近点（ICP）算法是点云刚体配准最常用的方法，但当2个点云偏移较大时可能会陷入局部极小。工件合理的初始位置可由人工操作来保证，但这种方式严重依赖于操作者的经验且精度有限，特别是对于毛坯件等缺少精确定位信息的工件就更加困难。本文提出一种基于边缘检测的预处理算法，利用少量边缘点计算最优的调整方向，将工件调整到合适的位置，以确保后续 ICP 方法的有效性。仿真结果表明，该方法能有效解决缺失点对 ICP 性能的影响，改善传感器覆盖率指标。
展开▼
