Title: Computational Steering Technologies based on Parameter Space Exploration
Name: Dr. Tobias Post
In order to obtain high quality manufacturing processes, these processes have to be controlled in a closed loop. In the presence of undesired defects, human experts have to decide how to adapt the process parameters in order to maintain the quality of a machining process. Simulations coupled with feature extraction techniques are used to find such parameter adaptations out of the space of all parameters to get an optimal production process. However, understanding and visualizing of such parameter spaces that describe process quality in dependence of hundreds of parameters is not achieved by the state of the art.
We will develop techniques for exploring the space of all parameters for the computational steering of simulations. An explicit goal of this project are methods that can simplify and visualize such parameter spaces and help experts to identify and adapt only relevant parameters. Information-theoretic concepts will be very useful to reduce the complexity of the underlying data sets and effectively summarize pertinent information. Techniques of large scale visual analytics and high-dimensional analysis will be necessary to deal with these problems. Naturally, comparative visualization will be a key ingredient for this research.
In this research project, we propose to develop techniques of computational steering of simulations that, in combination with appropriate visualization techniques, give process experts valuable additional insight into their processes to optimize them. The proposed research topic extends a number of existing research domains and combines these extensions to the envisaged visualization and computational steering environment for simulations.
Visual comparison of flow simulation data in a parameter space exploration for a virtual factory environment