Feature model-based grasp formulation

Abstract

The most flexible means of transferring components between machines within a work cell is provided by robotic manipulation. This implicitly requires that a stable grasp of each component can be determined. The CAD representation of the part may be used for this purpose. A conventional search over the solid model representation of the part for the optimal grasp can be computationally expensive, especially when non-planar surfaces are present. In this paper, we present a logical evaluation of the solid feature model of the component, that finds an acceptable grasp with little computational effort. While the procedure does not guarantee that a grasp will be found, its efficiency and success for the majority of components more than justify its use before an exhaustive search is conducted. The technique is illustrated by means of an example.