Automated assembly planning for three-dimensional mechanical products

Alan C Lin, Purdue University

Abstract

This research proposes a methodology for automatic generation of assembly plans for three-dimensional mechanical products. The primary goal is to develop an effective way to describe and plan for the assembly of mechanical products. The results can be used not only for manufacturing, but also as a feedback to the designer. The proposed methodology for assembly planning starts with the creation of solid models for the assembly design. Techniques of geometric reasoning are developed to infer mating and collision in formation from the assembly solid models. Information concerning non-geometric properties of assembly is also considered. A frame-based representation scheme is used to explicitly represent the non-geometric information. In generating assembly plans, a two-stage planning scheme is used. In the initial planning, assembly plans are generated based solely on product geometry considerations. A three-layer planning strategy has been developed which analyzes the part connectivity relationships in a hierarchical manner and ensures the insertion of individual parts free of collision. To make the assembly plans applicable to practical mechanical products, the initial assembly plans are further revised in the second stage under sequence constraints derived from non-geometric properties. Both high-level assembly sequence information and low-level assembly process details are included in the final assembly plans, allowing feasibility of application in different manufacturing environments. In addition to generating feasible assembly plans, the same methodology is applied to analyze the assemblability of a product during its design stage. The proposed planning strategy is graph-based and independent of the domain of assembly models. It can be extended easily to cover most mechanical products.

Degree

Ph.D.

Advisors

Chang, Purdue University.

Subject Area

Industrial engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS