Zusammenfassung
Robotic vision tasks require a large amount of annotated data for effective training and performance. However, the manual creation of such datasets is a complex and time-consuming process. Particularly for smaller companies or those with high product variability, there is a need to optimise the resources dedicated to creating datasets for machine vision applications. This paper investigates novel ways to acquire annotated vision data through the integration of collaborative robots. Two methods are outlined to streamline the generation of annotated datasets. One of these methods is then used to construct a dataset tailored for 6D robotic gripping of an automotive connector. The effectiveness of automated annotated dataset generation is demonstrated by training a deep learning model to accurately estimate the orientation of the automotive connector.
Nutzer