GENERATING AND EVALUATING STABLE ASSEMBLY SEQUENCES

High-level assembly planning systems generate plans for the automated assembly of mechanical products by robots. The sequences to be generat ed underlie several physical and geometrical constraints, and in addit ion have to be efficient to increase productivity. The challenges stil l facing the field are to develop efficient and robust analysis tools, and to develop planners capable of finding optimal or near-optimal se quences rather than just feasible sequences. The presented high level assembly planning system (High)LAP automatically considers physical an d geometrical constraints to generate and to evaluate stable assembly sequences. In this paper we propose a relational assembly model includ ing a CAD description and the specification of features and relations of the assembly components. We use an optional specification of an arb itrary hierarchy of assemblies to speed up and guide the generation of sequences. (High)LAP evaluates all feasible assembly sequences consid ering several criteria like separability and manipulability of the gen erated (sub)assemblies. Furthermore, the necessity of reorientation fo r a mating operation and parallelism during plan execution are conside red. Another important criterion is the stability of the generated (su b)assemblies. Most of the assembly planners developed up to date use h euristical or user-defined criteria to determine assembly stability fo r plan evaluation. The presented system is the first assembly planning system which automatically determines the range of all stable orienta tions of an assembly for plan evaluation. Therefore, we introduce a st ability metric and an algorithm to calculate all stable orientations o f an assembly considering friction. Experimental results are presented to demonstrate the efficiency of our assembly planning system.