Open Positions

The institute of combustion technology for aerospace engineering (IVLR) at the University of Stuttgart is an integrative unit of the DLR institute of combustion technology.  The institutes host approximately 80 scientists plus a number of students from various fields of specialization to address research question of modern gas turbines. Thermal conversion of sustainable energy carriers, e.g. green hydrogen, is a key element for the energy transition policy, while synthetic aviation fuels are without alternative for flights beyond mid-range distances. We support this development with the competence fields computer simulation, chemical kinetics and analytics, combustion diagnostics, mass spectrometry, multi-phase flow and high-pressure combustion. We develop and operate various laser diagnostic techniques to resolve the influence of turbulent flow on physical and chemical processes and to analyse the combustion process in detail. Quantitative measurement data form the basis for the development of innovative concepts, as well as for the verification and further development of numerical simulation models. In this context, current development activities should be supported by advanced data analysis. Over the course of time, we have improved our data analysis framework with state-of-the-art methods like ML-based computer vision. We use novel methods for building, training and validating our models. In this work, we aim to extent our well-established ML framework from object detection to tracking them over time. The basis for this development is formed by time-resolved Particle Image Velocimetry (PIV) and Shadowgraphy measurements. Not only does this facilitate to develop numerical models and deeper fundamental understanding of the evolutions of various objects during combustion, backward and forward propagation of the object identifiers and thereby enhancing detection and classification accuracy. Once established, the framework can be scaled to other measurement techniques.

Work Content

1. Familiarization with computer vision, neural networks for image processing
2. Understanding the data analysis pipeline for experimental data
3. Evaluation of currently available state-of-the-art object tracking methods
4. Advanced object tracking capabilities for scientific images
5. Implement the developed solution into the existing framework
6. Testing of the implementation

Qualifications

1. Scientific university studies, e.g., in the fields of computer science, physics, chemistry, process engineering, aerospace engineering, or mechanical engineering.
2. Strong programming skills, preferably in Python.
3. Experience in the field of computer vision and machine learning, enthusiasm for programming.
4. Fundamental understanding of combustion processes and spectroscopy is advantageous.
5. Teamwork, dedication, ability to work independently, and a scientific approach.

Application

If you are interested, please send your application documents (or questions) via email to basil.jose@ivlr.uni-stuttgart.de or use the given link. A complete application should consist of at least: brief motivational statement (max. ½ page), CV, certificates and transcripts.

The institute of combustion technology for aerospace engineering (IVLR) at the University of Stuttgart is an integrative unit of the DLR institute of combustion technology.  The institutes host approximately 80 scientists plus a number of students from various fields of specialization to address research question of modern gas turbines. Thermal conversion of sustainable energy carriers, e.g. green hydrogen, is a key element for the energy transition policy, while synthetic aviation fuels are without alternative for flights beyond mid-range distances. We support this development with the competence fields computer simulation, chemical kinetics and analytics, combustion diagnostics, mass spectrometry, multi-phase flow and high-pressure combustion. We develop and operate various laser diagnostic techniques to resolve the influence of turbulent flow on physical and chemical processes and to analyse the combustion process in detail. Quantitative measurement data form the basis for the development of innovative concepts, as well as for the verification and further development of numerical simulation models. In this context, current development activities should be supported by advanced data analysis. Over the course of time, we have improved our data analysis framework with state-of-the-art methods like ML-based computer vision. In this work, we aim to extent our well-established ML framework from physical space analysis to the frequency domain to enhance the precision, robustness and transferability of image segmentation to facilitate high-level statistical analysis. Wavelet, Fast Fourier Transform (FFT) and Inverse Fast Fourier Transform (IFFT) based filters are powerful tools for analyzing images, allowing us to extract important and recurring features and characteristics from complex scalar patterns. These techniques leverage the properties of the frequency domain to isolate and enhance specific aspects of the images. Once established, it can be scaled to ML-based methods like Fourier Neural Operators (FNO), etc. These serve to further improve our computer vision models for the data analysis of experimental images.

Work Content

1. Familiarization with computer vision, neural networks for image processing
2. Understanding FFT and IFFT filter based methods in image processing
3. Implement FFT based methods for the data analysis framework
4. Explore FFT-based methods in machine learning (eg. FNOs)
5. Compare the FFT-based methods with current methods
6. Testing of the implementation

Qualifications

1. Scientific university studies, e.g., in the fields of computer science, physics, chemistry, process engineering, aerospace engineering, or mechanical engineering.
2. Strong programming skills, preferably in Python.
3. Experience in the field of computer vision and machine learning, enthusiasm for programming.
4. Fundamental understanding of combustion processes and spectroscopy is advantageous.
5. Knowledge in machine learning frameworks like PyTorch is advantageous.
6. Teamwork, dedication, ability to work independently, and a scientific approach

Application

If you are interested, please send your application documents (or questions) via email to basil.jose@ivlr.uni-stuttgart.de or use the given link. A complete application should consist of at least: brief motivational statement (max. ½ page), CV, certificates and transcripts.