Projects

• AiEng
• Theme: An interdisciplinary, project-oriented degree program with an educational focus on artificial intelligence and engineering sciences
• Summary: The project focuses on the development of a cooperative bachelor degree program "AI Engineering" in Saxony-Anhalt. The interdisciplinary degree program will enable students to develop AI systems and services in the industrial environment and beyond and to accompany the associated engineering process - from problem analysis to commissioning and maintenance / servicing. At the Harz University of Applied Sciences, it focuses on the field of mobile systems and telematics. The project follows a student-centered didactic concept, which is supported by many practice-oriented (team) projects and a large offer of Open Educational Resources (OERs) with (e-)tutoring program.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Stefanie Krause
• Duration: 01-Dec-2021 – 30-Nov-2025
• Funding: Bundesministerium für Bildung und Forschung (BMBF)
• Partner(s): Universität Magdeburg; Hochschule Anhalt, Hochschule Magdeburg-Stendal, Hochschule Merseburg

• WeedAI
• Theme: Intelligent UAV-Based Weed Monitoring System for Selective and Site-Specific Herbicide Application
• Summary: The objective of the project is to develop an intelligent real-time monitoring and mapping system for the detection of weed distribution in cereal stands. For this purpose, high-resolution aerial image data is captured at low flight altitude and classified directly on the drone using optimised onboard AI image recognition during the overflight. The innovative procedure enables species-specific recognition at the level of individual plants. In this way, the project makes a significant contribution to reducing the use of pesticides in arable farming and thus promotes sustainable agriculture. The Harz University of Applied Sciences is significantly involved in several project modules and will take on the following tasks in particular: support in the construction of a UAV carrier platform with RGB carrier camera, development of software for flight planning and aerial flights, optimisation of image recognition based on the CNN model, training and testing of the CNN model with the existing image data, integration of image recognition into the UAV system, as well as the final validation of the overall WeedAI system. The project makes an important contribution to the Federal Government's National Strategy for Artificial Intelligence in the following fields of action: continuous strengthening of AI research as the basis for successful overall development in Germany; rapid transfer of AI research results into application.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Jing Liu
• Duration: 28-May-2021 – 27-May-2024
• Funding: Bundesministerium für Landwirtschaft und Ernährung (BLE)
• Partner(s): Leibniz-Institut für Agrartechnik und Bioökonomie e.V. (ATB), Bornim bei Potsdam; CiS GmbH, Bentwisch bei Rostock


• Best Buddy
• Theme: Best Buddy Intelligence
• Summary: The aim of the project is to use a smartphone or tablet as an input device for a robot companion, both computing power and sensor technology of the mobile device, in order to be able to control a complex, modular system. The robot is designed to intelligently interact with the user by detecting voice, face, and gestures, as well as surrounding objects. The existing robot system will be extended to include artificial intelligence through machine learning, object recognition and image processing techniques.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Kai Steckhan, M.Sc. Jerome Kuhle
• Duration: 01-Dec-2017 – 30-Apr-2020
• Funding: Bundesministerium für Wirtschaft und Energie (BMWI), Zentrales Innovationsprogramm Mittelstand (ZIM)
• Partner(s): Kinematics GmbH, Bernau bei Berlin


• CoRg
• Theme: Cognitive Reasoning
• Summary: Cognitive computing addresses problems characterised by ambiguity and uncertainty, meaning that it is used to handle problems humans are confronted with in everyday life. When developing a cognitive computing system which is supposed to act human-like we cannot rely on automated theorem proving techniques alone, because humans performing common-sense reasoning do not obey the rules of classical logics. This project aims at the construction of a cognitive computing system by modeling diverse aspects of human reasoning, by extending classical logical reasoning with non-monotonic reasoning like defeasible and normative logics in combination with machine learning. Different components for modeling the commonsense reasoning process will be developed and combined to a cognitive computing system which will be tested using benchmarks from commonsense reasoning. All this shall result in a cognitive reasoning system able to address problems which none of the techniques alone would have been able to address.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Sophie Siebert
• Duration: 01-Mar-2018 – 30-Apr-2021
• Funding: Deutsche Forschungsgemeinschaft (DFG)
• Partner(s): Prof. Dr. Ulrich Furbach, Dr. Claudia Schon, Universität Koblenz-Landau


• HarPer
• Theme: Harmony Perception
• Summary: The perception of consonance/dissonance of musical harmonies is strongly correlated with their periodicity. This can be shown by consistently applying results from psychophysics and neuroacoustics, namely that the just noticeable difference (JND) between pitches for humans is about 1% for the musically important low frequency range and that periodicities of complex chords can be detected in the human brain. Based thereon, the concepts of relative and logarithmic periodicity with smoothing can be introduced as measures of harmoniousness. In the auditory brainstem, the periodicity pitch frequency, which is not present in the input spectrum, occurs in addition in the response spectrum. To explain this, it can be argued that the most important factor during the neural processing is the highly non-linear transformation of the input signal into pulse trains (spikes) whose maximal amplitude is limited. This is can be demonstrated by a simple recurrent neural network model. In this PhD project, the frequency spectrum of the brain responses will be analyzed with electroencephalography (EEG), considering in particular musical dyads and triads as stimuli. Combining EEG with functional magnetic resonance imaging (fMRI), the corresponding activated brain regions will be localized and the coding principles in auditory brain areas will be investigated. The overall goal is to develop a model how the human brain perceives and processes musical sounds.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Maria Heinze
• Duration: 01.10.2017–30.11.2020
• Funding: FEM-Power-Projekt; Graduiertenstipendium des Landes Sachsen-Anhalt
• Partner(s): Prof. Dr. Rainer Goebel, Maastricht University


• Decorating
• Theme: DEep COnceptors for tempoRal dATa mINinG
• Summary: Today's production plants, homes, or businesses collect huge amounts of data. These data, e.g., recorded by sensor networks in a smart home, form a time series, often with additional spatial information. Interesting questions in this context are, how can regular and/or recurring behaviours be detected, and how can we identify deviations from expected behaviours. For modelling and prediction of spatio-temporal behaviour, approaches based on recurrent neural networks are an appropriate choice. Conceptors are a new neuro-computational mechanism which can be used to implement a variety of functionalities, like incremental learning of dynamical patterns, neural noise suppression, or top-down attention control in hierarchical online dynamical pattern recognition. The main approach to practically implement conceptors is using a recurrent neural network with a randomly created connectivity matrix as a dynamic memory or echo state networks. During the course of the research project, simple but powerful recurrent network models, especially networks based on conceptors, will be considered in combination with clustering methods. It shall be investigated not only how pattern recognition is possible in time series using these methods in general, but, wherever possible, as well their applications, e.g. object recognition with multicopters, the perception of harmony in music, etc.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): Dipl.-Inf. Falk Schmidsberger
• Duration: 01-Jan-2017 – 31-Dec-2018
• Funding: Deutscher Akademischer Austauschdienst (DAAD) im Programm des Projektbezogenen Personenaustauschs (PPP) mit Australien
• Partner(s): Prof. Dr. Oliver Obst, Western Sydney University


• TriOptScan
• Theme: Development of a mobile scanner for the detection of markings of different heights, depth, and contours
• Summary: The project TriOptScan aims at a mobile system, namely a hand-held scanner, which is developed for detecting and recognizing markings on workpieces. These markings have different heights, depths, and contours. To identify them, a combination of classical image processing and laser triangulation is used. In addition to that, 2D markings such as bar codes can be processed.The objective of the subproject at the Harz University of Applied Sciences is the development of algorithms in order to capture 2D and 3D image data with optical sensors and laser triangulation. The movements of the scanner over the workpiece are analyzed and corrected by means of suitable methods such as optical flow to enable evaluation of the image data.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Matthias Marks, Dipl.-Inf. Falk Schmidsberger
• Duration: 01-Apr-2015 – 31-Jul-2017
• Funding: Bundesministerium für Wirtschaft und Energie (BMWI), Zentrales Innovationsprogramm Mittelstand (ZIM), Kooperationsnetzwerk Assistenz in der Logistik
• Partner(s): Institut für Automatisierung und Informatik GmbH (IAI), Wernigerode; Loetec Elektronische Fertigungssysteme GmbH, Lutherstadt Wittenberg


• InspektoKopter
• Summary: During the course of the R&D project "InspektoKopter", a modular test system and a novel method for external inspection of wind turbine rotor blades will be developed using an unmanned air vehicle, which is a safe and reliable alternative to current manual testing. The starting point is an existing multicopter system, which will be optimized for the use on wind turbines. To inspect the structural condition of the wind turbine blades, the development of a suitable optical inspection procedure is planned using appropriate camera and sensor technology. In addition to this, an analysis software for the identification of damage patterns, interpretation of the sensor data and visualization and archiving of the "fitness" of rotor blades has to be implemented. Various detection methods shall help to recognize and localize surface damage (cracks ) or qualitative and quantitative material fatigue. The challenge here is the aggregation of individual sensor recordings consisting of image sequences and depth image scans, in an automatic workflow for creating a complex 3D model of the rotor blade under the particular scanning conditions. To ensure the reproducibility of the test, the automation of aerial surveys will be provided with the developed multi-sensor system which requires novel navigation technologies. A user-friendly software solution will be developed, that analyzes and annotates the optical measurement results and allows comparison of inspection data by some history mechanism.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): Dipl.-Inf. Falk Schmidsberger
• Duration: 01-May-2014 – 30-Apr-2016
• Funding: Bundesministerium für Wirtschaft und Energie (BMWI), Zentrales Innovationsprogramm Mittelstand (ZIM), Kooperationsnetzwerk InDiWa - Inspektion, Diagnose, Wartung
• Partner(s): Fraunhofer-Institut für Fabrikbetrieb und -automatisierung (IFF), Magdeburg; GEO-METRIK-Ingenieurgesellschaft mbH, Magdeburg; Bitmanagement Software GmbH, Berg near München


• RatioLog
• Theme: Rational Extensions of Logical Reasoning
• Summary: Human reasoning does not strictly follow the rules of classical logic. Explanations for this may be incomplete knowledge, incorrect beliefs, or inconsistent norms. From the very beginning of artificial intelligence (AI) research, there has been a strong emphasis on incorporating mechanisms for rationality into AI reasoning systems. Rationality cannot be restricted to cognitive tasks solely, but involves complex behavior and interaction with other subjects and the environment. This project aims at establishing a joint model for reasoning and behavior. For this, we will combine logical reasoning with the modeling of continuous systems by building on previous works on non-monotonic calculi and hybrid automata.We will extend classical logical reasoning by various forms of non-monotonic aspects, e.g. abduction or defeasible argumentation. This will not be done on a theoretical level only, but these extensions will be incorporated into the existing reasoning system E-KRHyper. The open domain question answering (QA) system LogAnswer that uses E-KRHyper and the free encyclopedia Wikipedia for answering natural-language questions will be turned into a system for rational question answering, which offers an excellent testbed for evaluating rational reasoning.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): Dr. Claus-Peter Wirth
• Duration: 04-Jun-2013 – 03-Dec-2015
• Funding: Deutsche Forschungsgemeinschaft (DFG)
• Partner(s): Prof. Dr. Ulrich Furbach, Institut für Informatik, Universität Koblenz-Landau


• InfraKopter
• Summary: This project aims at composing many infrared images, recorded with the help of flying robots, employing object recognition techniques. The use of flying robots (multicopters) offers several advantages compared to standard technologies (e.g. with plane or helicopter). Multicopters are less expensive in acquisition and maintenance and are quickly ready for operation. Moreover images can be taken at relatively low heights (< 150 m), which allows us higher quality images to be taken. Since conventional methods of image processing (e.g. edge detection) on infrared images appear to be less suitable because of blurry contours, methods of photogrammetry and semantic image analysis (object recognition) and methods of artificial intelligence, in particular from machine learning, shall be applied in this project in order to generate stitched images. Application areas include agriculture and forestry or archaeology.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): M.Sc. Matthias Marks
• Duration: 01-Jan-2013 – 31-May-2015
• Funding: Bundesministerium für Wirtschaft und Energie (BMWI), Zentrales Innovationsprogramm Mittelstand (ZIM)
• Partner(s): GEO-METRIK-Ingenieurgesellschaft mbH, Magdeburg


• AirMeter
• Summary: This project deals with the development of a universal sensor platform for image and environmental data detection as well as of a flight assistant system for semi-autonomous radio-controlled flight systems (multicopters). The use of flying robots (multicopters) offers several advantages compared to standard technologies (e.g. with plane or helicopter). Multicopters are less expensive in acquisition and maintenance and are quickly ready for operation. Moreover environmental data detection and image acquisition can be taken at relatively low heights (< 150 m) and close to objects of study (< 3 m).
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): Dipl.-Inf. Falk Schmidsberger
• Duration: 01-Nov-2011 – 30-Apr-2014
• Funding: Bundesministerium für Wirtschaft und Energie (BMWI), Zentrales Innovationsprogramm Mittelstand (ZIM)
• Partner(s): Geomatics – Ingenieurburo für angewandte Informationstechnologie, Wernigerode


• DeMAS
• Theme: Deductive design, analyses and verification of multiagent systems in the RoboCup
• Summary: Specifying behaviors for collaborative (physical) multiagent systems and multi­robot systems is a sophisticated and demanding task. Due to the high complexity of the interactions among agents and the dynamics of the environment the need for precise modeling arises. Formal design and verification methods barely exists. In this project, a method will be enveloped to design the software of such systems in a formal matter. In our approach, these designs are directly executable specifications. Therefore we use methods from deduction and logic-programing (Prolog) and the Unified Modeling Language (especially statecharts). With this approach, it is possible to verify and to analyze multiagent systems. For these tasks, temporal and dynamic logics and model checking methods will be used and developed further. By this, a system design is possible, which can be integrated in multi­robot systems. The concrete integration will be realized in a RoboCup simulation team and a real robot team (Type Sony Aibo). Finally, we will have a system which allows the design, implementation, formal analysis, and verification of collaborative multi-agent systems and mobile robots. In cooperation with other teams in the DFG research programme, the usage of our method in related approaches and system architectures is intended.
• Project Leader: Prof. Dr. Frieder Stolzenburg
• Staff Member(s): Dipl.-Inf. Falk Schmidsberger
• Duration: 01-Jul-2001 – 31-Dec-2008
• Funding: Deutsche Forschungsgemeinschaft (DFG) im DFG-Schwerpunktprogramm 1125 RoboCup (Kooperierende Teams mobiler Roboter in dynamischen Umgebungen)
• Partner(s): Prof. Dr. Ulrich Furbach, Institut für Informatik, Universität Koblenz-Landau

Publications

• Chronological List
• Google Scholar