Content-based CAD assembly model retrieval: Survey and future challenges

Currently, the content-based retrieval is a problem of major interest in several different fields and, focusing on mechanical engineering, many approaches exist to compare and retrieve single CAD parts, to evaluate shape similarity, to extract features and to segment models. However, most of the proposed approaches do not take into account all the key characteristics of an assembly model, such as the relationships between its components, and the different levels according to which two assembly models can be considered similar, i.e. either globally, partially, or locally. For these reasons, the retrieval of CAD assembly models still faces challenges to fully satisfy designers' expectations. The aim of this paper is to review the state-of-the-art of works addressing the CAD assembly model retrieval and to identify future challenges and possible research directions. Firstly, the paper highlights the user requirements for CAD assembly model retrieval and proposes a set of criteria for analyzing the available methods grouped into the following macro-categories: objective, assembly characterization, assembly descriptor, query specification and type of similarity. Secondly, it describes and characterizes the available methods by organizing them according to the adopted criteria. Finally, it discusses the open issues and future challenges.