AI and Robotics for GeoEnvironmental Modeling and Monitoring (AIRGEMM)
Digitalization and Artificial Intelligence (AI) are increasingly pervading all areas of science, industry, society, and everyday life. The aim of the project AI and Robotics for GeoEnvironmental Modeling and Monitoring (AIRGEMM) is to develop scientific foundations for the use of AI within the focus areas of the TU Bergakademie Freiberg in earth, material, energy and environment sciences. Of particular interest are the development of AI methods and applications in the areas of environmental robotics and modeling of geological structures and processes as well as theoretical contributions to machine learning and algorithms.
The aim of the robotic boat sub-project is to develop an autonomous platform that serves as a carrier for various sensor systems for 3D mapping and environmental data acquisition in inland waters. For this purpose, a robotic boat is augmented with navigation sensors and self-driving skills for autonomous operation in inland waters. The robotic boat serves as carrier for various sensor systems for environmental mapping and monitoring. 3D point clouds obtained from ultrasonic and LiDAR mapping are semantically segmented and classified through machine learning analysis. To prepare an application in Saxon inland waters (e.g. lakes, river dams), the AI algorithms are trained with synthetic data obtained in virtual environments.
The geo-modeling sub-project applies machine learning methods on complex geophysical data sets. Scientific and technical goals concern e.g. the development of AI-based prediction methods for the detection of disturbances and boundary layers. These predictions will form the basis for significantly more efficient simulations, e.g. for solving of inverse problems in the area of electromagnetic geophysics. Further work addresses e.g. automated data processing workflows to establish model-driven machine learning pipelines using heterogenous geological data repositories.
The project runs from 01.07.2019 to 31.12.2021. Four institutes and eight professorships are involved in the project. For more information please see Research Team.
The central objectives of the project AIRGEMM and the partner project RoBiMo (see Partner Project) are:
- 3D depth-resolved acquisition of inland water quality parameters with an autonomously moving swimming robot
- Development of new sensors for the determination of nitrate and microplastic contents
- Validation of the results with in situ measurements as well as sampling by scientific divers to perform further analytical methods
- geoscientific evaluation of water quality parameters and respiratory flows for a better understanding of limnic processes
- Visualization of the data with methods of artificial intelligence and virtual reality
The envisioned Deep Learning applications of AIRGEMM induce a massive demand on available compute and data processing power. To accommodate these needs, the Institute of Computer Science operates an AI supercomputer. The configuration of the NVIDIA DGX-2 server is as follows:
- GPUs 16 x NVIDIA Tesla V100
- GPU Memory 512GB total
- CPU Dual Intel Xeon Platinum 8168, 2.7 GHz, 24-cores
- Performance 2 petaFLOPS
- RAM 1.5TB
- Network 2 x 10Gbit/s fiber optics
- Storage 30TB
- Prof. Dr. Konrad Froitzheim (Institut für Informatik)
- Prof. Dr. Christian Gerhards (Institut für Geophysik und Geoinformatik)
- Prof. Dr. Heinrich Jasper (Institut für Informatik)
- Prof. Dr.-Ing. Bernhard Jung (Institut für Informatik)
- Prof. Dr.-Ing. Sebastian Zug (Institut für Informatik)
- Prof. Dr. Oliver Rheinbach (Inst. für Numerische Mathematik und Optimierung)
- Prof. Dr. Ingo Schiermeyer (Institut für Diskrete Mathematik und Algebra)
- Prof. Dr. Klaus Spitzer (Institut für Geophysik und Geoinformatik)
- Gero Licht, M. Sc. (Institut für Informatik)
- Mandeep Kaur, M.Sc. (Institut für Diskrete Mathematik und Algebra)
- Samuel Kost, Dipl.-Math. (Institut für Numerische Mathematik und Optimierung)
- Armel Perod Nya, M. Sc. (Institut für Informatik)
- Stefan Reitmann, Dr.-Ing. (Institut für Informatik)
Jarosch, L.; Pose, S.: Reitmann, S.;Dreier, O.; Licht, G.; Röder, E.: Roboter für das Wasser der Zukunft, Acamonta, 2020, Freiberg, Deutschland
Kost, S.; Rheinbach, O; Schaeben, H. Using logistic regression model selection towards interpretable machine learning in mineral prospectivity modeling, Geochemistry, Volume 81, Issue 4, 2021. https://doi.org/10.1016/j.chemer.2021.125826
Pose,S.; Reitmann,S.; Jarosch,L.; Dreier,O.; Röder,E.; Licht,G.: Automatisiertes Gewässermonitoring, Fachbeitrag, wwt wasserwirtschaft wassertechnik Nr. 04/2020, Berlin, Deutschland. PDF
Pose, S.; Reitmann, S.; Licht, G.; Grab, T.; Fieback, T.: RoBiMo –The tasks of scientific divers for robot-assisted fresh-water monitoring, Freiberg Online Geoscience, Vol. 58, ISSN1434-7512, S. 32-38, June 2021.
Reitmann, S; Jung, B. Generating Synthetic Labeled Data of Animated Fish Swarms in 3D Worlds with Particle Systems and Virtual Sound Wave Sensors. 2nd International Conference on Cyber-Physical Systems and Control, CPS&C’2021. To appear. (*Best Paper Award)
Reitmann, S.; Kudryashova, E.; Jung, B; Reitmann, V. Observation Stability and Convergence for Neural-type Evolutionary Variational Inequalities. Differential Equations and Control Processes, Issue 2, 2021. ISSN 1817-2172
Reitmann, S.; Kudryashova, E.; Jung, B; Reitmann, V. Classification of Point Clouds with Neural Networks and Continuum-Type Memories. Proc. 17th International Conference on Artificial Intelligence Applications and Innovations - AIAI 2021. Springer, S. 505–517. https://doi.org/10.1007/978-3-030-79150-6_40
Reitmann, S.; Neumann, L.; Jung, B. BLAINDER — A Blender AI Add-On for Generation of Semantically Labeled Depth-Sensing Data. Sensors 2021, 21, 2144. https://doi.org/10.3390/s21062144
Richter, F.; Reitmann, S. & Jung, B. Integration of Open Geo Data into Virtual Worlds. In: Proc. 6th International Conference on Virtual and Augmented Reality Simulations (ICVARS 2022).
Weit, S; Börner, R.-U.; Brändel, M; Gödickmeier, P.; Gootjes, R.; Kost, S; Rheinbach, O.; Scheunert, M.; Spitzer, K. Convolutional Neural Networks Applied to 2D and 3D DC Resistivity Inversion. In J. H. Börner & P. Yogeshwar (Eds.): Proceedings of the Schmucker-Weidelt Colloquium on Electromagnetics, 27. - 30. September 2021. Deutsche Geophysikalische Gesellschaft, ISSN 2190-7021.
This measure is co-financed with tax revenues on the basis of the budget decided by the Saxon state parliament, grant application № 100376434 to the State Ministry for Higher Education, Research and the Arts (SMWK) of the Federal State of Saxony.
The work of the AIRGEMM project is closely linked to the sister project of the ESF junior research group RoBiMo on the topic of robot-assisted inland water monitoring. This junior research group is supported by the State of Saxony with funds from the European Social Fund until the end of December 2022.
Institut für Informatik
Telefon +49 (3731) 39 39 39
frzinformatik [dot] tu-freiberg [dot] de