基于形状先验模型的平面型工件抓取位姿检测

发布时间：2018-01-22 01:46

本文关键词： 工件分割形状先验抓取位姿结构光视觉系统　出处：《机器人》2017年01期 　论文类型：期刊论文

【摘要】：为了能够在复杂的工业环境中抓取平面型工件,提出一种图割法与形状先验模型结合的工件图像分割方法,并且对工件的位姿信息进行测量.首先,建立先验形状,提出基于最小包围矩形法将工件的形状模板与目标工件人工分割形状进行配准,得到先验形状.为了保证分割结果的准确性,采用单一的先验形状.图割模型中加入了目标形状先验知识.其次,通过自适应调整形状先验项的权重系数,使得图割法的能量函数形状先验项自适应于被分割的图像.第三,本文可以采用形状先验方法分割一幅图像中的多个工件,并且能够计算吸盘的最优抓取位置.最后,采用结构光视觉系统采集工件的点云信息,拟合工件平面,确定工件法向量,得到工件的抓取姿态.实验结果表明,本文算法能够适应遮挡、光照变化的工件图像,同时也能够分割复杂环境中的目标工件;平面型工件抓取位姿的计算结果有效,可以应用于遮挡、反光、复杂干扰背景条件下的工件抓取作业.
[Abstract]:In order to be able to grab a planar workpiece in the complex industrial environment, put forward a segmentation method combined with graph cut method and shape prior model of the workpiece image, and measure the position information of the workpiece. First of all, the establishment of a priori shape, puts forward the minimum bounding rectangle method based on shape template and target workpiece workpiece shape manual segmentation registration, get the prior shape. In order to ensure the accuracy of the segmentation results using a single prior shape. Graph cut model is added in the target shape prior knowledge. Secondly, the weight coefficient adaptive shape priors of the graph cut method of energy function shape prior adaptive in image segmentation. By third, we can the shape prior segmentation method of multiple workpieces in the image, and can calculate the optimal grasping position of sucker. Finally, collecting workpieces with a structured light vision system The point cloud information, fitting the workpiece plane, determine the normal vector, get the workpiece grasp posture. The experimental results show that this algorithm can adapt to occlusion, illumination changes of the workpiece image, but also can separate the workpiece target in complex environment; calculation of planar grasping pose result is effective and can be applied to occlusion. Reflective, grasping operation and complex background interference conditions.

【作者单位】：中国科学院大学;中国科学院自动化研究所;
【基金】：国家863计划(2013AA041002-1) 国家科技支撑计划(2015BAK06B01) 国家自然科学基金(61403372,61403374)
【分类号】：TP242;TP391.41
【正文快照】： 1引言(Introduction)工业机器人抓取作业要求准确识别平面型工件的位姿,确定吸盘的抓取姿态.工件的位姿确定从两个方面考虑,首先是工件的图像分割,确定其位置信息;其次是平面型工件的姿态检测,从而确定吸盘的姿态.平面型工件是指一类具有平整表面的工件.但是工业现场的工件存

【相似文献】