中文摘要：

机器人在空间环境下完成对复杂形状物体的精细操作,要求其具有较强的自主操作规划能力.从学习人类经验出发,模仿人类经验生成来建立经验数据库,提出有效的物体分类方法,在物体类别基础上,识别同类物体的可抓取区域,并提出＂单拇指＂姿态规划方法,将对物体的操作经验泛化到同类物体的操作规划中.试验证明了算法的有效性.

英文摘要：

It is a challenging problem for space robots autonomously to accomplish precise manipulation on objects with complex geometry. In this paper, an effective method for object classification and graspable components identification is firstly proposed by imitating the generation of human experience. Then, a ＂single＂ finger grasp planning method is developed which is able to generalize the grasp to the objects in the same category. Experiments verify the effectiveness of the proposed methods.