The research project addresses social challenges in the area of tension between demographic change and digitization. The "baby boomer" generation accounts for around one fifth of the total population in Germany. As they gradually enter retirement age, municipalities are challenged to create age-appropriate care structures. The goal is to develop a system that serves public stakeholders as a decision-making aid in their planning processes using methods from artificial intelligence, software development and mathematical optimization. For this purpose, the Optimization working group (Mathematics) creates and examines complex models from route and location planning, among other things.

The project is funded by the Carl-Zeiss-Stiftung.

Homepage

Goal of SynphOnie (Synergies from physical and traffic planning models for multi-criteria optimization of multi-modal demand-oriented transport)

Given a mobility demand and a road network, the task is to determine a "good" transportation offering.

"Good" means:

• Short travel times for passengers
• Low costs for operators

• Low CO₂ emissions, low energy consumption

   

Specific Questions:

• Is a tram needed?
• In which areas do we provide scheduled bus services? 
  Which routes are established?
• Where do we complement with ridepooling offers? 
  How many vehicles are needed?

• For which mobility desires do we recommend a car?

   

Homepage

Homepage

Homepage

AI-Care: Artifcial Intelligence for treating cancer therapy resistance

Problem Formulation

Cancer is one of the leading causes of death worldwide and has recently surpassed cardiovascular disease as the number one cause of death in high-income countries. Glioblastoma is an especially belligerent brain tumor known for its notorious behavior of adapting its cellular machinery in ways that allow it to become unresponsive to treatment. It has an incidence of 0.59 to 5 per 100,000 people per year. The average life expectancy after diagnosis is around 17 months due to recurrence. Therapeutic failure has multiple causes, but the two most important underlying factors are the heterogeneity of the genetic makeup of glioblastoma and the instability of the cancer cell regulatory processes at the genetic and epigenetic level. The diverse phenotypic traits that result from the combined effect of these microscopic factors exhibit high plasticity at the macroscopic level, which further contributes to therapy resistance. Plasticity here refers to the ability of cells to transition from one phenotypic state to another due to external and/or internal microenvironmental factors. This is typically observed as a response to drug treatment, rendering tumors likely to switch to a more resistant state.
 

Solution Approach

Faced with the poor understanding of the plasticity phenomenon in general, the heterogeneity of patient-specifc phenotypic plasticity profiles and the high dimensionality of the associated data, the use of AI methods is crucial to tackle the complexity characterizing the plasticity landscape and to advance towards individualized treatments against a resilient tumor. Given the diversity of cellular states driven by the malignant transformation and the increased plasticity, the current datasets generally represent very sparse temporal snapshots with little predictive power regarding the temporal evolution of the tumor. In our project, we tackle these challenges via two approaches. Firstly, we augment the datasets with synthetic data using datadriven style transfer techniques and generative models, such as GANs (generative adversarial networks), to enhance the representation of sparse cellular states. Secondly, we develop multiscale models based on hybrid AI modeling, which captures both discrete states and continuous trajectories by leveraging autoencoders and optimal transport approaches. Multi-modal experimental datasets including transcriptomic, epigenomic and proteomic data will be combined in integrative AI models, in which they will be projected into a common phenotypic space. In such a reduced space, dimensions correspond to phenotype features (signaling and metabolic pathway activities corresponding to cancer hallmarks and master transcription factor activities) of the normal and malignant cells. The projection into a lower-dimensional space, whose dimensions have a biological interpretation, is based on interpretable AI models, incorporating a priori biological knowledge in the form of biological graphs within an autoencoder structure. These models will be enhanced by applying state-of-the-art methods such as self-supervised learning (contrastive learning or masked autoencoders) to increase their phenotypic resolution.
 

Research Questions

In light of above facts, the three most important research questions of this project are:

• How can we characterize an interpretable phenotypic space to delineate the phenomenon of tumor plasticity in glioblastoma?
• What is the survival strategy that a cancer cell applies and how does it relate to the pathway selection and motion dynamics in the phenotypic landscape?
• How can we use AI-models for devising groundbreaking personalized glioblastoma therapy strategies?

Homepage

In ever smaller pipelines, the wall properties have an increasing influence on the flow conditions. Thus, with the aid of applied coatings consisting of microstructures, wall properties and their flow influence can be brought about in a targeted manner. Air pockets enclosed within these functional structures reduce, for example, the flow resistance of a liquid flowing over them, according to the lotus effect. Application perspectives are, among others, miniaturized processes or microsystems. In contrast to previous concepts, the surfaces developed within the project promise better applicability as well as higher stability of the enclosed air. The research combines analytical modelling, numerical simulation, surface fabrication and experiments. The partner of the project is the Photonik-Zentrum Kaiserslautern.

This project is funded by the Deutsche Forschungsgemeinschaft (DFG).

Machine learning methods are being used in more and more areas of our world. There are also opportunities for industry, for example in the optimization of products or the monitoring of industrial processes. Chemical processes and their plants are highly complex and produce huge amounts of heterogeneous time series data that are recorded by various sensors at different measuring points. Even the most experienced plant operators cannot keep track of all this. Safety is the top priority. If the process runs into a critical area, it must be shut down in time if necessary. However, this is usually very expensive. Unnecessary shutdowns must therefore be avoided. The task now is to identify critical conditions early and reliably. Humans alone cannot do this. Partners of this project are the TU München and the Universität Oldenburg.

This project in funded by the Deutsche Forschungsgemeinschaft (DFG).

EASIER (sEAmless SustaInable EveRyday urban mobility) is an interdisciplinary, transnational research project, funded for 3 years. The goal of EASIER is to increase the share of active and sustainable personal transport modes, such as walking, biking, public transport and shared mobility services, and to integrate them to appealing multimodal journeys. The subproject „Efficient tariff systems for sustainable mobility“ at RPTU aims to use tariff systems actively to increase the demand and share of public transport. Therefore, tariff systems and pricing strategies for integrated mobility systems are analyzed and developed, focussing on passenger behavior, their route choice and the underlying line network.

The project is funded by the Bundesministerium für Bildung und Forschung (BMBF) via ERA-NET Cofund Urban Accessibility and Connectivity (ENUAC).

Homepage

The meniscus regeneration and involved cell and tissue-level phenomena pose a difficult biomedical problem. Clinical studies indicate that partial and total meniscectomies lead to prevalence of premature osteoarthritis in knee joints. Therefore, substantial efforts are being made towards finding adequate regenerative tissue for meniscus replacement. Mathematical modelling, simulation and experimental validation can provide basic control mechanisms in cell-scaffold interactions under different environmental parameters, and yield a selective prognosis of the most significant combinations of these parameters. For the application of in-silico modelling and simulation a challenging part is the well-posed and numerically efficient coupling of the processes at the cell level with the macroscopic behaviour and the mechanical properties of the tissue. Partners of the projekt are the DITF Denkendorf and the Universität Ulm.

The project is funded by the DFG as part of the program SSP 2311.

The HALF project enters the next round. The goal of the BMBF pilot innovation project "Energy-efficient AI systems" was the development of energy-efficient hardware that enables artificial intelligence for the evaluation of cardiac arrhythmias on mobile devices. For this purpose, a cross-layer approach was developed in the HALF project in collaboration with Fraunhofer ITWM, which enables a Pareto-optimal hardware implementation including an automated neural network search. The project was awarded 1st place in the FPGA category in this competition by the Federal Minister of Education and Research in March 2021, and was awarded a new BMBF project in which a new ECG method for the automated mobile cardiac arrhythmia detection was developed, which has a 5x longer runtime with simultaneously improved detection accuracy compared to the state of the art.

This project is funded by the  Bundesministerium für Bildung und Forschung (BMBF).

Homepage

The innovation platform KEEN connects 20 industrial and scientific institutions with the aim of introducing artificial intelligence technologies and methods in the process industry. As part of KEEN, the research team is working on the development of hybrid predictive substance data models. Substance data are of fundamental importance in process engineering, but since their experimental determination is expensive and time-consuming, predictive methods are needed. While physical methods are well established for this purpose, data-driven machine learning methods open up completely new perspectives. In the project, both worlds are brought together and superior hybrid approaches, which combine the expressivity of machine learning methods with the extensive existing physical knowledge, are developed.

The project is funded by the Bundesministerium für Wirtschaft und Energie  (BMWi).

Homepage

Wood and other cellulose fiber insulating materials are the most commonly used insulating materials made from renewable resources. However, their thermal conductivity is generally higher than the thermal conductivity of conventional insulation materials. Due to the manufacturing process, the distribution and orientation of the cellulose fibers lead to highly anisotropic thermal conductivity. Besides single fibers, the microstructure also contains fiber bundles of different sizes. To this end, accurate prediction of thermal conductivity as well as further optimization of the board structure to achieve thermal conductivities <35 W/K is difficult. The goal of the research project is therefore to optimize the structure of highly porous wood fiber insulation boards in order to further reduce their effective thermal conductivity. The potential for this lies precisely in exploiting anisotropy and specifically mixing fiber bundles of different sizes. To exploit this potential, machine learning methods, image based geometric structure modeling and numerical methods for the efficient simulation of heat transfer are to be combined with optimization methods. Partner of this project are the Fraunhofer ITWM and the Bergische Universität Wuppertal.

This project is funded by the Bundesministerium für Bildung und Forschung  (BMBF).

Quantum technologies are pushing ever more strongly and rapidly into industrial application. The development is so fast in some areas that the demand for specialists cannot be met by graduates of universities. To enable working professionals to qualify in quantum technologies at an academic level, the RPTU Kaiserslautern-Landau (RPTU) will develop a part-time, interdisciplinary distance learning course. The German Federal Ministry of Education and Research (BMBF) is providing around two million euros over three years for the conception and development of this course as part of the funding programs for quantum technologies. Under the leadership of Prof. Artur Widera, we are working in mathematics together with colleagues from the departments of electrical engineering and information technology, computer science, and physics, as well as the Fraunhofer ITWM and the DFKI, on new teaching content and formats to enable optimized distance learning at the Distant and Independent Studies Center (DISC) at RPTU. The new master's program, called QuanTUK, will be launched in three years at the Distance and Independent Studies Center (DISC) of the RPTU. It will be located at the interface between physics, mathematics, computer science, and electrical engineering and information technology.

This project is funded by the Bundesministerium für Bildung und Forschung  (BMBF).

The connection of CAD methods and finite elements was established in the last decade by isogeometric analysis (IGA for short). Here, the calculation areas are often defined by their edges. This is exactly where Scaled-Boundary Isogeometric Analysis (SB-IGA) comes in, which deals with the parameterization of domains based on NURBS boundary surfaces. The project, which is carried out in cooperation with the Chair of Structural Dynamics at RWTH Aachen University (Prof. Sven Klinkel), focuses on the coupling of different computational domains as well as the global smoothness of the underlying approach functions. Furthermore, the advantages of the SB-IGA approach in the context of "trimming", i.e. adding and removing regions in CAD, will be investigated. The theoretical considerations and methods will then be applied in the field of structural dynamics.

This project in funded by the Deutsche Forschungsgemeinschaft (DFG).

Machine learning is an artificial intelligence process that is currently in the process of changing almost all areas of our lives: from personal assistants to medicine and health to self-driving cars. ML systems are capable of extracting knowledge from large amounts of data, but they require high computational power and a lot of energy to do so.

The goal of this project is to develop methods that reduce the energy consumption of this technology so that it can also be applied to smaller control units of machines. To this end, basic machine learning methods are to be modified in such a way that the background knowledge of humans can be used to require less data and less computing power.  The project is coordinated by Professor Dr. Paul Lukowicz from the Department of Computer Science at the RPTU. The project partner is the German Research Center for Artificial Intelligence.

The project is funded by the Carl-Zeiss-Stiftung.