A topological model of multifingered prehension

Abstract

The authors present a topological model of prehension for multifingered robot hands which essentially deduces low-level contact wrench parameters in joint space from high-level task specifications, object properties, and a topological space (tetrahedron) of hand shapes. This model is based on a theory of prehensility with control algorithms modeled after the functionalities and sensory-motor activities of the human hand. The authors describe representations at different levels of granularity. They have sketched topological equivalences between high-level and low-level spaces, which preserve three topological properties: power, precision, and support (or equivalently, graspability, manipulability, and supportability). They introduce the notion of functionality distribution of power, precision, and support to a set of hand subconfigurations. They note the subtle difference in support as an independent parameter (from power and precision), and support as a dependent parameter (to strengthen power and precision) and identify general schemes to handle the difference. The model offers a framework in which several existing schemes may be integrated.<>