The research department of Agents and Simulated Reality links innovative basic research with the application oriented implementation and transfer of solutions to industry.
We therefore combine research from Artificial Intelligence, High Performance Computing, Visual Computing and Security into solutions in the fields of Autonomous Driving, Industrie 4.0, Smart Environments and Computational Sciences.
The ability to accurately model and simulate reality and thus make predictions is a key element of our fast-moving industry. The Agents and Simulated Reality Department uses dynamic large-scale, highly detailed 3D environments to realistically model relevant scenarios, such as complex production lines, ancient cities, or biological cells. Adding semantic information provides the context to run simulations that accurately predict properties (e.g. illumination, acoustics, traffic, etc.) of such systems. High-performance implementations on multi- and many-core systems enable highly realistic realtime display, e.g. using Realtime Ray Tracing, as well as realtime interaction, even with complex simulated realities using fully immersive VR environments.
Modern societies strongly depend on traffic and transport. Latest developments in the field of driver assistance systems like lane or break assist point towards a future where vehicle operation will be increasingly supported if not fully taken over by computer systems. Thus making travel not only more comfortable but also way more safe than today. In order to make the vision of fully autonomous driving a reality the computer systems have to reliably asses critical traffic situations and act accordingly. The main obstacle here is that critical situations do not occur too frequently in real life (fortunately) and cannot be reproduced easily in order to obtain a sufficiently large data sets to train algorithms. The aim of the research department of ASR within the field of Autonomous Driving is to come up with simulations that generate training data sets for critical traffic situations.
To create valid scientific models of things like cells or materials is crucial to gain knowledge about their internal structure. These models are used to predict behavior and properties. Modern 3D microscopy systems and other 3D imaging systems like magnetic resonance imaging produce large amounts of image data from the examined probes. They create 3D images of the structure’s internal elements, their spatial relation, their number and size. In order to identify these elements the image data has to be filtered and analyzed. Due to the high amount of noise in the images produced by some imaging techniques, excessive image processing is necessary. High Performance Computing techniques help to decrease the computing times for large sets of 3D images. Finally methods from the field of Artificial Intelligence are employed to detect the proper elements in the images.
An application example of the research is the simulation and visualization of production processes: for the increasing individualization of products and shorter product cycles, potential cost reduction lies within the optimization of setup and turnaround times. With multi-agent systems, the processes of individual process steps are recorded and simulated in 3D. The formal modeling of systems and their processes as hybrid systems allows guaranteed statements, which at the same time take into account the discrete control software as the spatio-temporal behavior of plants. By running a visual inspection of the facilities and their operations based on the three-dimensional representation of the simulation and verification, a rapid assessment of the planning is possible. Interactive and immersive training scenarios on a virtual 3D model can be made during the conversion of the plant.
Today’s live is made more easy and more comfortable at the same time more energy efficient by various technical means. Ranging from the personal smartphones, to home automation to modern energy grids. From Smart Homes, to Smart Cities and Smart Energy grids the scale varies. Although every one of them presents a technological and societal breakthrough in themselves the full potential of these Smart Environments could only be harvested when systems on all levels of the scale work together. In the long term every environment will consist of a multitude of Autonomes Systems working together providing individual case-based support. Artificial Intelligence is the classical research area for such autonomous cognitive systems that are able to perceive their environment, learn and derive plans and act accordingly.
High Performance Computing
High Performance Computing is the key technology for break through innovations in many fields. Visual Computing, Big Data analysis, Machine Learning or Simulation to just name a few among many others. Most modern application scenarios either deal with the processing of large amounts of data e.g. from sensors or work with large amounts of data like simulations. While the underlying algorithms of the software systems dealing with these vast amounts of data are rather stable their actual implementation depends on the given hardware they are implemented on. And hardware changes in very short cycles. This makes the efficient transfer of algorithms to new hardware platforms a key factor in leveraging the potential of new hardware developments.