Feature reasoning for automatic robotic assembly and machining in polyhedral representation†

Abstract

An automatic planning system for robotic assembly must have knowledge of the workspace. In particular it must be capable of reasoning about the shape of objects within the workspace. This requires a knowledge of object features and their behaviour in assembly situations. In this paper, a system is presented which extracts features relevant to assembly and machining from a boundary model derived from a Constructive Solid Geometry representation of a complex shape. In particular, finding concavities plays an important role in identifying possible mating features for assembly. These relevant features can then be used for assembly planning based on finding compatible sets of mating features. In this work the CAD data is extracted from a boundary model represented in an ASCII file and converted into the data-structures of the AI language, Prolog.