Datag love based grasp planning for multi-fingered robot hand

J Liu,YR Zhang
2004-01-01
Abstract:We developed a dataglove based grasp planning system for grasp simulation, visualization, and analysis of motion mapping from human hand to robot hand. Mapping in joint coordinate system during preliminary grasp. is achieved. The system can effectively analyze motion mapping from human hand to robot hand in a virtual environment. It is also useful in building virtual prototypes of different hand design and determine how the design affect the grasp ability. This paper describes the working principle, the hardware and software architecture, and the grasp strategy. Furthermore, we developed a classifier for human hand grasp according to features of human grasp represented by Gaussian Mixture Model (GMM). The taxonomy of robot hand grasp can be achieved by the joint coordinate mapping from human hand to robot hand. Our aim is to control robotic finger motions through human grasp experience.
What problem does this paper attempt to address?