Gierecker, JohannJohannGiereckerKalscheuer, FlorianFlorianKalscheuerSchoepflin, DanielDanielSchoepflinSchüppstuhl, ThorstenThorstenSchüppstuhl2023-10-092023-10-092023Conference 16th CIRP Conference on Intelligent Computation in Manufacturing Engineering - Procedia CIRP 118: 930-934 (2023)https://hdl.handle.net/11420/43586Industrial manual assembly, especially within the area of large-scale assembly, often lacks process monitoring since the feedback about performed assembly tasks is only generated by the worker. Optical sensor systems thereby offer the potential to monitor assembly states automatically and derive the required information without intervening the work in progress. With the growing trend of customization, not only the assembly process itself, but also the monitoring system must be adapt- able at short notice. However, most machine vision systems, as they are today, are commonly task-specific solutions and are therefore hard to be transferred to another inspection task or other work objects. To lower the barriers on applying machine vision into varying environments, this paper introduces an automated CAD-based sensor planning and implementation pipeline. An analysis and derivation of common constraints in assembly design is laid out, followed by a method of generating and optimizing inspection features and sensor poses. A strategy to implement the image processing pipeline based on the derived features is presented and applied on an assembly use case.en2212-82712023930934Elsevier B.V.https://creativecommons.org/licenses/by-nc-nd/4.0/assemblyflexible automationindustry 4.0vision systemssensor planningTechnologyConference Paper10.15480/882.868510.1016/j.procir.2023.06.16010.15480/882.8685Conference Paper