Return to search

Automated grasping for articulated structures using evolutionary learning algorithms.

The human face and hands, as well as the animation thereof, are very complex and are difficult to model realistically in computer graphics. A lot of research, that deals with the different aspects involved in modelling the idiosyncrasies of the face and hand, has been devoted to the accomplishment of this task. In this dissertation, we will focus on the human hand, one of the most complex and useful tools humans have at their disposal. The hand and its agility are attributes which differentiate us from most living mammals. Previous papers focusing on the human hand have dealt with the construction and animation of realistic looking human hands, the stability of a hand grasp and automatic grasping techniques. Papers on forward and inverse kinematics of articulated structures provide background knowledge useful in animating the hand. We want to accomplish the automated grasping of an arbitrary object by our hand model using evolutionary algorithms to determine the final placement of the hand and fingers, instead of telling each finger where to touch or grasp the object. The program should allow the hand to learn by itself where to place each finger, based on heuristic information that is supplied by the program. To successfully imitate the grasping hand in a computer graphics environment we have to investigate the restrictions placed on motion by the skeletal structure of the hand, the effect of the palm, the deformation of the skin and tissue, as well as attributes of the hand and objects to be grasped. Physical properties of the object such as the strength and weight of the material and physical forces influence the grasp. What adds further to the complexity of animating the hand, is the intricate design and structure of the hand itself: The great number of degrees of freedom enables the hand to adapt to a large number of goal specific configurations, where the hand can appear as a flat surface, rounded fist, etc. / Hardy, A.

Identiferoai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uj/uj:2419
Date29 May 2008
CreatorsGlobisch, Ralf
Source SetsSouth African National ETD Portal
Detected LanguageEnglish
TypeThesis

Page generated in 0.0014 seconds