IRVINE, the research outcome of this collaboration, is an interactive data science system to detect and understand previously unknown errors in the manufacturing process of electrical engines at BMW. IRVINE incorporates concepts and techniques from Machine Learning and Artificial Intelligence, as well as Information Visualization and Visual Analytics.
With IRVINE, engineers at BMW can now explore, detect and label previously unknown errors in the manufacturing process of electrical engines faster and more efficiently. Since the deployment of the system, numerous new error signatures could be identified and characterized.
As a result, a considerable number of engines and engine parts could already be prevented from scrapping, due a systematic early detection of errors.
The paper is a model design study on applying visual analytics to a manufacturing process and makes strong contributions to the interactive data science community towards an enhanced human-machine collaboration for a practical, emerging, and largely unexplored data analysis problem. With the focus on electrical engines, this work also contributes to more sustainability in manufacturing, as well as to ecological responsibility.
The Best Paper Award at IEEE Vis (~top 1%) recognizes outstanding work from the pool of accepted papers based on various criteria, including the potential impact to the community, the importance of any results obtained, and technical challenges to be overcome.