AN EFFICIENT SYSTEM FOR GEOMETRIC ASSEMBLY SEQUENCE GENERATION AND EVALUATION

In this paper, we present a software system which can automatically determine how to assemble a product from its parts, given only a geometric description of the assembly. Incorporated into a larger CAD tool, this system, the Stanford Assembly Analysis Tool (STAAT), could thus provide immediate feedback to a team of product designers about the complexity of assembling the product being designed. This would be particularly useful in complex assemblies where each designer may not be fully aware of the impact of his design changes on the assemblability of the product as a whole. STAAT’s underlying data structure is an efficient version of the non-directional blocking graph (NDBG), a compact representation of the blocking relationships in an assembly. STAAT implements several techniques using this structure, under a unified approach in which the same software “machinery ” can analyze the product under different assembly constraints. In initial experiments conducted on relatively small polyhedral assemblies of 20 to 40 parts and 500 to 1500 faces, using one-step translational motions, STAAT generated assembly sequences much more quickly than did previous NDBG-based systems. We are working now on extending both these results and the underlying theory to more sophisticated cases