Towards Automatic Extraction of Product and Process Data for Human-Robot Collaborative Assembly

Abstract

Automatic data extraction from CAD files needs to handle the increasing product individualization and variety. This paper reports on first results from a research project that uses a pipeline to extract data dealing with a high product variety for human-robot collaborative assembly. The approach extracts information from a CAD model, a 2D drawing and an assembly instruction list and subsequently merges and stores the information in a data model. A novelty of our approach is the merging of different information sources to automatically generate a new CAD model. We demonstrate and evaluate our approach using an industrial use case which has multiple product variants. The usage of the extracted data is the generation of assembly sequences for human-robot collaboration. Our preliminary conclusion is that the extraction of product and process data can be made more flexible and adaptable by utilizing multiple information sources.