There is a trend for a holistic perspective in engineering optimization based upon knowledge generation from large heterogeneous data sets, machine learning and search algorithms. New types of human user interaction enable a learning-driven adaptation of the system to not only come up with co-created innovative solutions but also to allow the system to innovate itself online.
At HRI-EU, we research on key components like efficient representations ranging from traditional CAE methods and shape morphing techniques for reasonable shape variations to modern geometric deep learning approaches for building compact encodings in an unsupervised fashion.
Machine learning allows us to capture the essence from data sets and build models for fast optimization circles in complex industrial applications, leading to the first notion of experience transfer. In addition, human user interaction recorded as process knowledge builds upon the engineer’s engagementin the current task for extracting user preferences and self-adapting the optimization system. This bridges the gap from a pure support system to a partnering system in a shared Decision Making environment. This second notion of experience utilization follows the idea of generating a co-creative environment for a cooperative workforce to strive for innovative solutions.