摘要
机器人多指手的抓持分类是抓持规划的一个主要问题。提出了一种基于支持向量机和二叉树聚类的分类方法来实现机器人多指手的抓持模式分类。根据多指手预抓取物体的几何特征,对机器人多指手的预抓取模式进行识别与分类,仿真结果表明该方法能快速而有效地实现机器人多指手的抓取模式分类。
Pre-grasp classihcatlon of multiple fingers is one of the key problems in grasp planning of Robots,This paper expounds a method based on support vector machine and binary tree with clustering to elassfy the pre-grasp pattern of multipul fingers in terms of the geometric features of objects. The experiment result shows that this method can give the pre-grasp classification of multiple fingers rapidly and effectively.
出处
《机械工程与自动化》
2006年第4期94-96,共3页
Mechanical Engineering & Automation
基金
江苏省应用基础研究课题(BJ98057)
南京航空航天大学科研创新基金(Y0487-031)
关键词
多指手
预抓取模式
支持向量机
分类
multiple fingers
pattern of pre-grasp
support vector machine
classification