On multiple interpretations

In the literature on features, there has been a wall between feature recognition and process planning. Much of the manufacturing knowledge, which is typically used in process planning, is rarely incorporated into feature recognition. After the output of a feature recognizer is fed into a process planner, there is little communication between these two activities. This paper proposes to integrate feature recognition and process planning, and presents an effort toward this integration in the context of multiple interpretations. A set of features required to create a part is called a feature model or an interpretation of the part. A part can have multiple interpretations. This paper classifies existing approaches for multiple interpretations into two schools, analyzes the complexities of their algorithms, and reveals that the nature of multiple interpretations is combinatorial. Therefore, algorithms that try to generate all interpretations or an optimal interpretation are subject to combinatorial explosion. As a solution, this paper presents a feature recognizer that computes a satisficing interpretation and generates alternative interpretations on request from a process planner.