Model Based Definition for Robotic Assembly

Abstract

This thesis deals with the product information requirements associated with automated robot programming for assembly operations. The first contribution of this thesis is the establishment of information flow from design to robotic assembly using STEP AP242Ed2 files. Model Based Definition (MBD) methodology is followed to capture the product manufacturing information (PMI) along with the product geometry during the design phase. MBD is the backbone of the Digital Thread (DT) in a connected enterprise. MBD is created using the second version of Application Protocol 242 of ISO 10303 standard released in 2020, also known as STEP AP242Ed2. PMI is added to the STEP AP242 files semantically by two methods. One way of adding the PMI is by using the standard entities defined in the STEP standard. Another method is using custom-defined Unicode strings in alignment with the industry-recommended practices. Using these two methods, almost all the PMI traditionally included in the 2D manufacturing drawings can be semantically added to the STEP AP242 files. Another contribution of this work is extracting and reusing the product information, both geometric and PMI, for robotic task specification. The spatial relationships between the constituent parts in an assembly, GD&T, welding information, and surface finish annotations are semantically added to the STEP AP242 files. A motor assembly is used as a test case to illustrate the extraction of spatial relationships and their subsequent application in defining manipulator motion constraints for robotic assembly. Welding symbols and GD&T are semantically embedded within the STEP AP242 files, facilitating the extraction of relevant PMI and associated part features. This extracted data, ranging from worst-case boundaries of mating features to welding annotations, plays a crucial role in robotic assembly operations, from estimating assembly forces to specifying welding constraints. Furthermore, this thesis introduces a novel classification of handling features and criteria for gripper selection, leveraging product data from MBD files. The surface finish, threading, material, and other information extracted from the semantic annotations are instrumental in selecting grippers and planning grasping and handling tasks. The practicality of this approach is underscored through assembly experiments, where product data from MBD files is harnessed for robot programming. This thesis contributes to the two significant areas of Industry 4.0, the establishment of DT and the adaptation of smart automation using robotic manipulators. Adopting the MBD methodology using STEP AP242Ed2 files facilitates the DT in the extended enterprise, including the suppliers and automatic constraint definition for robotic task specification with minimum human intervention