Using Near-Misses From Feature Recognition to Generate Redesign Suggestions for Increased Manufacturability

Abstract

This paper presents a method for generating design alternatives aimed at making a mechanical part more manufacturable. The method is intended to be used in the context of a human-computer collaboration in which the computer suggests design alternatives and the designer decides which alternatives effectively meet his or her overall design goals. The mechanism presented in this work for producing design alternatives uses “near-misses” generated during the feature extraction process. Most approaches to suggestion generation produce design alternatives through use of library of alternatives. However, this is not a very general or flexible approach. The approach proposed in this paper offers a more general method that can replace the library method. Advantages include a more general mechanism for generating design alternatives, and generation of a broader variety of more contextually appropriate alternatives.