机器人多指手预抓取模式聚类分析

提出一种基于计算机视觉的,以及改进的模糊C均值聚类算法的机器人多指手预抓取模式分类方法.根据人手抓取分类学,将抓取手势分为13类.选取若干具有代表性的不规则形状物体,先经视觉系统采集物体图像,然后运用数字图像处理方法提取物体的姿态、大小、形状和表面粗糙度等特征,最后利用改进后的模糊C均值聚类算法对待抓取物体进行聚类分析.实验结果表明:对比人类抓取策略,该方法具有理想的预抓取模式分类效果.

Pre- grasp Pattern Clustering Analysis for Multi- fingered Hands

A pre- grasp pattern classification method based on the computer vision and the improved fuzzy C - means clustering algorithm is introduced. According to the grasp taxonomy, the grasp hand shapes is divided into 13 types. Some representative irregularly shaped objects are selected for the research. First of all,images are acquired through the computer vision system, and then using the digital image processing methods, the characteristics of each object, including the posture characteristic, the size characteristic, the shape characteristic,the surface roughness characteristic, and so on, are extracted. Finally, by the means of the improved fuzzy C- means clustering algorithm, the pre - grasp patterns are classified for the objects to be grasped. The experiment results of the method indicate that in contrast with the human grasp tactics, the method obtains the ideal pregrasp pattern classification.