A multi-agent approach to fixture design

Abstract

The design of fixtures is a highly complex process that requires the human designer to draw from his rich experience. In addition, for a given workpiece, multiple solutions may exist. By exploiting the recent advances in CAD/CAM and Artificial Intelligence techniques, one may constrain the multiple solutions such that only good designs (measured through performance measures) are considered. In this paper, a multi-agent fixture design system is proposed that harnesses the advantages of genetic algorithms and neural networks. This system attempts to capture the relevant domain knowledge and uses it to produce acceptable solutions efficiently. The system is applied to a case problem and the suggested fixturing solution is compared to one offered by a human designer. The agreement between the two solutions is very close.