Förderprojekte
Die Forschungstätigkeit am Lehrstuhl wird primär durch Drittmittel finanziert. Der Lehrstuhl ist vernetzt mit einer Vielzahl von nationalen und internationalen Forschungs-, universitären und Industriepartner. Die Projekte adressieren Aufgabenstellungen der modell- und datenbasierten Regelung, Inferenz und Entscheidungsfindung in komplexen technischen und biologischen dynamischen Systeme in Bereichen:
- Cyber-physische Systeme: Modellierung, Regelung und Resilienz
- Automotive: Autonomes Fahren, Fahrdynamikregelung und Elektromobilität
- Robotik: Kooperative mobile und stationäre Robotik
- Energietechnik: Wirtschaftliche Konzepte, Sektorkopplung und Demand-Side Management
- Produktionstechnik: Digitale Zwillinge, Optimierung, Regelung und Monitoring
- Eingebettete Systeme: Ereignisbasierte Protokolle und Regelung
- Systembiologie: Genome und Populationsmodelle vom Krebs und Viren
- Verfahrenstechnik: Modellierung und Regelung in der Kristalisations- und Granulatiostechnik.
Die am Lehrstuhl laufende bzw. kürzlich abgeschlossene Förderprojekte werden im nachfolgenden Gantt-Diagramm grafisch dargelegt. Die gesamte Fördesumme beträgt ca. 10 Millionen Euro. Weitere Details zu den jeweiligen Projekten werden weiter unten angegeben.
AI-Care: Artificial Intelligence for Treating Cancer Therapy Resistance
The AI-Care project, a collaborative effort between RPTU, DKFZ, and the University of Heidelberg, aims to unravel the fundamental principles underlying cancer cell plasticity and devise personalized therapeutic strategies that overcome the emergence of resistance. Glioblastomas are aggressive brain tumors characterized by a high degree of phenotypic heterogeneity and plasticity. Their ability to switch to resistant cell states renders conventional therapies ineffective. The goal of this project is to decipher the cancer plasticity code of glioblastoma using state-of-the-art AI and optimization techniques that characterize the drug response dynamics of glioblastoma.
Funded by: Carl-Zeiss-Stiftung
Time span: January 2024 - December 2029
AIMPID: AI-based Mutation Predictions and Relevant Protein Inhibitor Development in SARS-CoV-2
AIMPID is focused on predicting mutations in SARS-CoV-2 genome and developing small inhibitor molecules which target the virulent protein products of predicted mutations to block their activities. This helps in controlling the spread of pandemic as frequent mutations in SARS-CoV-2 genome (especially in case of targeted RNA vaccines) are taking place, rendering the developed drugs useless. Since SARS-CoV-2 is an RNA virus, it is prone to mutations; therefore, it necessitates the tracing of mutation patterns in the viral genome to find the least mutable regions in order to design enduring and more stable inhibitor molecules. The main tasks constituting the conduction of the whole project are mutation rate prediction, mutation prediction and development of small inhibitor molecules against proteins translated from mutated RNA sequences.
Funded by: Bundesministerium für Wirtschaft und Klimaschutz
Time span: June 2021 - May 2023
DEEPCOR: Deep Reinforcement Learning based Prediction of SARS-CoV-2 Behavior
DeepCor analyzes the strategies followed by SARS-CoV-2 for its survival and evolution. The idea is based on the concepts of exploration and exploitation from reinforcement learning (RL), where the virus acts as the agent and tries to survive in the environment by taking some actions. The focus of the work is on the genome of SARS-CoV-2 that how certain genomic changes help the virus to adapt and survive. The RL algorithms in DeepCor are based on policy gradient methods, actor critic methods or Q learning methods. DeepCor algorithms will help explore the survival policies opted by the virus in response to environmental stresses. The second part of this project focuses on the rates at which the virus changes its behavior, which is done by the help of recurrent neural networks (RNNs). The rates will help in identifying the hotspot strategies that are necessary for the evolution of virus. The end goal of the DeepCor is to provide novel insights in h-SARS-CoV-2 behaviors such that effective medicinal targets can be recognized.
Funded by: Ministerium der Wissenschaft Rheinland-Pfalz
Time span: August 2021 - December 2023
Adaptive data-driven predictive control using behavioral approach for autonomous powder compaction
The overall aim is to develop and implement a data-driven control algorithm for autonomous powder compaction on a rotary tablet press. This includes an online autonomous adjustment of quality attributes such as dose and hardness, while minimizing the waste during start-up, maximizing the production rate and compensating the process disturbances during production. Our interpretation of the autonomous process behavior implies an online and adequate self-adaptation of the setpoints of the process parameters such as punch distance, impeller speed, etc. in attempting to maintain the product quality. The latter may indeed undergo various deteriorations as a result of the impact of inherent process disturbances, such as material flow fluctuations or varying material properties (e.g. particle size). The objective thereby consists in enhancing and improving the product quality and process effciency in comparison to the manual process management, which has been a common practice in the pharmaceutical industry today.
Funded by: Deutsche Forschungsgemeinschaft
Time span: January 2023 - December 2025
Autonomous control of a process chain for CO2 carbonation by use of mine waste
The aim is to develop the individual process steps for CO2 storage and the production of recyclable carbonates and to link them to form an autonomous process chain. To this end, a Self-Learning Robust Adaptive Control(ler) (SLARC) algorithm, consisting of three interconnected parametric statistical estimators, is introduced. The estimators are primarily meant to be implemented via (deep) neural networks (DNN) and thus can be trained using data and process knowledge (model). SLARC is to be understood as a kind of hybrid model which fuses physical laws with the information gathered from data. The former makes it possible to detect unfavorable/unphysical states by constraining to physics of the process while the latter makes it possible to detect changes in particle properties during operation and to adapt to the current state of the process. These two combined with the reinforced online learning feature make SLARC a self-learning algorithm which is also robust to random disturbances.
Funded by: Deutsche Forschungsgemeinschaft
Time span: January 2023 - December 2025
AORTA: Automatisierte Bildung von Rettungsgassen in komplexen Szenarien durch intelligente Vernetzung
Ziel von AORTA ist es, durch die automatisierte Bildung der Rettungsgasse einen Beitrag zum automatisierten Fahren sowie engen, technischen Vernetzung zu leisten, Verkehrsunfälle mit schweren Verletzungen bis hin zur Todesfolge zu verhindern, Einsatzfahrzeuge schneller und sicherer an ihr Ziel kommen zu lassen und somit täglich Leben zu retten! Erreicht wird dies mittels einer Integration von Infrastruktur, Sensorik, Kommunikation, HMI und Fahrzeugtechnik welche koordinierte Entscheidungsebenen verschiedener Abstraktionsgrade von der Einsatzleitstelle bis hin zum automatisierten Fahrmanöver auf klein- bzw. großflächigem Raum ermöglicht. Dabei wird eine auf Optimierung und Fahrzeugkommunikation basierende, dezentrale Plattform, die kooperative Fahraufgaben zum Bilden einer Rettungsgasse mittels zentraler oder/und verteilter Entscheidungsfindung durchführt, entwickelt werden.
Funded by: Bundesministerium für Verkehr und digitale Infrastruktur
Time span: January 2021 - December 2023
RADSPOT: HIGHLY AUTOMATED AND ROAD-FRIENDLY DRIVING BASED ON GROUND PENETRATING RADAR SIGNALS
This project focuses on the development of innovative AI-based autonomous driving algorithms boosted by ontological and knowledge graph models living in a Digital Twin (DT). To this end, a framework of hierarchical reinforcement learning, consisting of a multi-layer decision policy is applied. The learning agent(s) is (are) then able to choose not only elementary actions, but also to learn how to combine missions at a higher abstraction level. On the other hand, the DT maps the physics of the entire traffic on a road segment and, in its current development stage, serves as a cloud-based predictive maintenance of the road infra- and substructure. E.g. novel road-preserving autonomous driving can be implemented, by taking into account the inferred damage spots during a real-time path planning.
Funded by: Bundesministerium für Verkehr und digitale Infrastruktur
Time span: October 2018 − December 2021
ZEBRA: ZEB- und Georadar- Datenerfassung sowie kombinierte Zustandsbewertung von kommunalen Straßen mittels automatisierter KI-Analysemethoden
Als Nachfolgeprojekt von RADSPOT, realisiert ZEBRA die Kombination von ZEB und Georadar-Bewertung auf kommunaler Ebene am Beispiel der Stadt Kaiserslautern. Mittels KI und Digitalisierung wird ein Cloud-Dienst entwickelt, der die Zustandsbewertung und das Monitoring des Aufbaus- und der Oberfläche von Straßen ermöglicht. Die entwickelte Lösung soll mittelfristig auf weitere Kommunen erweitert werden.
Funded by: Bundesministerium für Verkehr und digitale Infrastruktur
Time span: January 2023 − December 2024
VASCO: KI-basiertes Visuelles Assistenzsystem zur Computergestützten Orientierungs- und Handlungsfähigkeit für Menschen mit Seheinschränkungen
For the assistance of visually impaired people, a wearable assistance system will be developed in the VASCO project, which consists of input devices (camera, microphone), a computing unit, and output devices (headphones, tactile devices) for the transmission of information to the user. The combination of specially developed energy-efficient computing technology, including sensor and actuator technology and fast, computationally efficient AI algorithms, gives the assistance system unique characteristics in terms of ergonomics and functionality range.
Funded by: Bundesministerium für Bildung und Forschung
Time span: January 2023 − December 2025
Weiterlesen
eCells: Automatisierte Flexibilitätsaggregation und zelluläre Energiemanagementsysteme für hochdigitalisierte Netzinfrastrukturen und Verteilnetze
Das Ziel des Projekts besteht darin, eine digitale Plattform zu entwickeln, welche eine optimale, autonome und dynamische Bündelung von verteilten Flexibilitäten (Lasten, Erzeugungsanlagen und Speichern) auf der Ebene der Mittel- und Niederspanung ermöglicht. Hierzu soll eine hochdigitalisierte Netzinfrastruktur über Edge und Cloud verteilte Softwareplattform geschaffen werden, welche die Medienbrüche zwischen der Sensorik, den zur Berechnung notwendigen Algorithmen und den anzusteurnden Aktoren überwindet, um in Echtzeit automatisch und sicher auf Flexibilitätsanforderungen reagieren zu können. Eingesetzt werden dabei anspruchsvolle prädiktive und datenbasierte Algorithmen der Regelungstechnik und KI. Das Projekt umfasst drei Anwendungsfälle (virtuelle Kraftwerke, Fläxibilitätsaggregation, Energiegemeinschaften) und verfolgt das übergeordnete Ziel mittelfristig einen wichtigen Beitrag zur Sicherstellung des Netzbetriebes und zum kostensenkenden Netzausbauprozess zu leisten.
Funded by: Bundesministerium für Wirtschaft und Klimaschutz
Time span: January 2024 − December 2026
MultiLOAD: Ressourcenschonendes Batteriemanagement für bidirektionales Laden unter Einsatz einer Modularen Multilevel-Konverter-Topologie
Das Förderprojekt "MultiLOAD" adressiert Potentiale der Ladearchitektur und Mechanismen des sog. “Modularen-Multilevel-Converter” (MMC) mit besonderem Bezug auf BDL. MMC-Module können dynamisch, während des Betriebs zu- und abgeschaltet werden. Dadurch können die jeweiligen Module bestehend aus einer Serien- und Parallelschaltung von Zellen in allen Betriebsmodi dynamisch geschaltet werden. In diesem Kontext sollen in MultiLOAD optimierte schaltende Ansteuerverfahren für MMC entwickelt werden, um eine verminderte zyklische Alterung der Zellen unter Berücksichtigung der BDL-Belastung zu erzielen.
Funded by: Bundesministerium für Wirtschaft und Klimaschutz
Time span: April 2023 − March 2026
DeRIVE: Netzdienliches und bidirektionales Laden an der Schnittstelle zwischen Energie - und Verkehrssektor
The aim of the project is to develop electromobility solutions for electric vehicle fleets that are integrated into energy networks in order to offer new network-related services. These should help in addressing the challenges of simultaneously minimising the necessary electric grid expansion and reducing the total cost of electric vehicles using cleverly designed incentive models. For this purpose, technical solutions for users, operators and energy providers including building administrations, shops and parking space managers are provided.
Funded by: Bundesministerium für Verkehr und digitale Infrastruktur
Time span: July 2022 − June 2025
DESPRIMA: Demand-Side- und Produktions-Management für Getränkeabfüllprozesse
DESPRIMA addresses modeling and control tasks in energy efficient DSM based production. In particular, we develop, validate and practically implement hierarchical and distributed planning and control strategies of production and energy management for both, individual and coupled production lines. To this end, nonlinear dynamic models for beverage filling machines and production lines shall be first developed. The resulting holistic models describe the dynamics of the production machines, the impact of the power flexibility potential, as well as the DSM system services and energy consumption constraints. The DSM-based energy-efficient planning and control of the individual and coupled production lines adapt in real time to changes in the environment, such as energy consumption, production requirements, weather conditions, stock market and process disruptions. The control algorithms comprise a broad design framework covering MPC, energy-efficient hierarchical production and a hybrid dynamical setup.
Funded by: Bundesministerium für Wirtschaft und Energie
Time span: July 2019 − June 2022
VirMan: Virtual Manufacturing Plant - Model and Data-Driven Predictive Maintenance
Digital twins are a virtual representation of a production system synchronized between the virtual and the real system, able to run different analytics based on sensed data and smart devices, optimized real-time communication protocols, and mathematical models. Industry 4.0 manufacturing paradigms exploit these features to forecast and optimize the behavior of the production system at each life cycle phase in real-time. Keeping interoperability and modularity at the core, VirMan project aims at the development of structural and functional semantic-based information models in order to build a digital twin for the virtual plant. Further tasks equip it with novel and real-time connectivity infrastructure and connected smart devices, setting up the data storage and processing environment, developing AI modules to analyze a large amount of data and developing plant monitoring and predictive maintenance functionality, and data visualization modules. The project outcomes include interoperable software and hardware modules for digital twin of a manufacturing plant based on ISA 95 and Industry 4.0 standards. TUK develops data-driven control, condition-monitoring and predictive maintenance algorithms in a use-case with the large water production enterprise Brandenburger Ustromquelle.
Funded by: EUREKABundesministerium für Verkehr und digitale Infrastruktur
Time span: May 2021 - April 2023
ReMiX: Resilienz in Mixed-Criticality-Systemen des Industriellen Internet der Dinge
In the ReMiX project, a design methodology for verifiable system architectures in intelligent automation is to be developed. For this purpose, distributed resources are summarized as a shared virtual resource and organized according to the principles of mixed-criticality systems. Mixed-Criticality describes a mapping of functions to resources based on their criticality according to available resource quotas. The distributed resources are merged as a shared virtual resource and organized according to the principles of systems with different criticality. The research results of this project will contribute to increase the system resilience through new design methods for self-organizing communication, computing and control approaches. By integrating security aspects into the design methodology, we aim to extend our development framework to attack-resistant mixed-criticality systems.
Funded by: Bundesministerium für Bildung und Forschung
Time span: September 2019 − August 2022
KIMKO: Multifunktionale mobile Roboterplattform für ein digitales Produktionsfeld der additiven Fertigung
The aim of KIMKO is to develop a robot system consisting of a mobile platform, two lightweight robots and stereo cameras for use in a 3D printing farm. The main research topic within this project is the collisionfree online trajectory planning for the manipulators and the mobile platform as well as their coordination in order to navigate autonomosly and to cooperatively plan and perform the robot motions/tasks in a confined place. Due to the high structural flexibility, model-based predictive control strategies are used for trajectories generation, whereas AI-based machine vision methods are beeing considered for the environment perception and online map generation.
Funded by: Bundesministerium für Wirtschaft und Energie
Time span: August 2019 − December 2021
CooPick: Kollaborative Roboter-Roboter-Mensch Interaktion beim Fruchtauflegen
CooPick develops a flexible and scalable robot system, which can be integrated into existing manufacturing processes and is able to perform task support in a cooperative robot-robot and robot-human interaction. Several lightweight robots communicate with each other and coordinate their actions and missions. Therefore, centralised and distributive collision-free model predictive control (MPC) algorithms combined in a hierarchy with scheduling have been developed. In a use-case a fruit-sorting task has been considered, where additionally an interplay with human operators is addressed.
Funded by: Bundesministerium für Wirtschaft und Energie
Time span: January 2018 − June 2020
KORINS: Komplementäres Robotergestütztes Inspektionssystem für Mehrwegflaschen
Im Bereich der Getränkeindustrie wurde nach der Einführung von einheitlichen Glasflaschen sowie sogenannten Modulkisten eine Verschärfung von Individuell-Marketing zur Vernachlässigung der einheitlichen Glasflaschen und Kisten deutlich ausgeprägt und umgesetzt. Bei der Rückgabe der leeren Pfandflaschen kommt es zu großen Vermischungen von unterschiedlichen Flaschentypen und Individualflaschen in den Leergutkisten. Einige Hersteller führen daher eine manuelle Nachinspektion durch, damit immer mehr Flaschen dem Pfandpool erhalten bleiben. KORINS entwickelt ein automatisches Sortiersystem, welches in bestehende Systeme integrierbar ist und basierend auf KI-Methoden über zwei Stationen die Flaschen und Kisten inspiziert und sortiert.
Funded by: Bundesministerium für Wirtschaft und Energie
Time span: September 2020 − August 2022