AN INTEGRATION OF NEURAL-NETWORK AND RULE-BASED SYSTEMS FOR DESIGN AND PLANNING OF MECHANICAL ASSEMBLIES

Currently, a case associative assembly planning system (CAAPS), which integrates neural computing techniques and rule-based systems has been developed. The neural network computing captures the designer's design concept and self-organizes similar experienced designs. The CBAPM (CLIPS-based assembly planning module), a component of CAAPS, generates a task-level assembly plan automatically. The design concept is expressed by a standard pattern format representing components' three-dimensional (3-D) geometry. A feature-based model translates the conceptual design into a preliminary boundary representation (B-rep). Based on a refinement of the B-rep assembly representation, assembly plans are generated for practical use in a single-robot assembly work-cell. A feasible assembly plan that minimizes tool changes and subassembly reorientations is generated from the system. In contrast with many assembly planning systems that used a prolonged question-and-answer session or required knowledge beyond what is typically available in the design database, the CBCAPM presented here draws input relationships directly from the conceptual design and the geometry of the assembly. The purpose of the CAAPS is to provide the engineer with an environment in which he/she can think of assembly in terms of high-level features and synthesize such assembly rapidly. At all stages of the design process he/she can consult the design cluster memory (DCM) and plan cluster memory (PCM) to see what ''experience'' knows of similar assemblies. Efficient use of prior experiences is emphasized. The designed DCM and PCM can dynamically continue to accumulate experiences. The developed algorithms can recall entire assembly plans or devise new plans on demand for a given new task in the design and planning environments. Several experiments illustrate the effectiveness of the designed assembly planning system.