On the concurrent design of assembly sequences and fixture

In this dissertation, I address the problem of automatically generating both an assembly sequence for a product and a fixture to hold all the intermediate subassemblies, given only a geometric description of the product. The fixture must stabilize every subassembly against gravity and against part-on-part insertion forces, without blocking any of the part motions. Unlike in previous assembly-planning work, the sequence and fixture are generated concurrently rather than sequentially. I start by presenting an efficient system for the assembly-sequencing problem alone, which I developed jointly with other researchers. This system utilizes geometric-reasoning techniques to generate one or more assembly sequences and to analyze the complexity of the assembly, as measured by various figures of merit. The introduction of fixturing to the problem improves the realism of the analysis, and allows physically unstable sequences to be eliminated from consideration. The joint problem of designing an assembly fixture concurrently with the sequence raises a number of new issues, which I explore. I present two methods to solve this problem. The first method is fast and successful in most (but not all) cases. The second method, based on the new concept of the insertion-force stability cone, is complete under certain assumptions, and is useful primarily when parts are assumed to be "glued' (or similarly attached) in place upon insertion. Both methods have been implemented for the case of planar assemblies, and experimental results are presented for each. The long-term vision for this work is a computer-aided design tool which can provide immediate feedback to a product designer about the difficulty of assembling the proposed product, without the need to construct a prototype.