The aim of this thesis is to build a MLOps-based system to compare the variations in the motions executed while performing various sports-related movements. In essence, we are interested in comparing movements across multiple sensor-suit based recording sessions.
Thesis Type |
|
Student |
Chenhuan Gao |
Status |
Finished |
Presentation room |
Seminar room I5 6202 |
Supervisor(s) |
Stefan Decker |
Advisor(s) |
Michal Slupczynski |
Contact |
slupczynski@dbis.rwth-aachen.de |
In our currently running project MILKI-PSY, we focus on multimodal immersive mentoring to facilitate Self Regulated Learning (SRL) of psychomotor skills. To support mentors with the tools they need to provide adaptive and personalized tutoring, a set of learning services was developed and is currently under development. These (micro-) services are built in accordance with modern development principles.
Recent developments in wearable sensors, augmented reality, artificial intelligence, and machine learning which allows modern computer systems to provide contextual support for learning psychomotor skills such as performing different sports. This allows to make use of a cross-domain approach to gather multimodal recordings of expert activities and to use these recordings to provide contextual feedback for learners. However, comparing complex data streams that result from such recordings across different body types and movement capabilities requires a complex algorithmic foundation.
To provide mentors and learners with a detection mechanism to help detect and understand differences in motion executions, the goal of this thesis is to build a framework capable of ingesting and comparing multiple motion data streams resulting from recording different actors performing a number of movements.
Potentially relevant literature:
- Björn Krüger et al. 2010: Fast Local and Global Similarity Searches in Large Motion Capture Databases
https://doi.org/10.2312/SCA/SCA10/001-010 - Serhan Gül et al. 2020: Kalman Filter-based Head Motion Prediction for Cloud-based Mixed Reality
https://doi.org/10.1145/3394171.3413699 - Paaßen et al. 2022: Teaching psychomotor skills using machine learning for error detection
https://ceur-ws.org/Vol-2979/paper1.pdf
If you are interested in this thesis, a related topic or have additional questions, please do not hesitate to send a message to slupczynski@dbis.rwth-aachen.de.
Please apply with a meaningful CV and a recent transcript of your academic performance.
- Must:
- Python
- Kuberentes, Docker
- Web technologies
- Beneficial:
- Machine Learning
- DevOps
- Data analytics / visualization
MILKI-PSY - Multimodales Immersives Lernen mit KI für Psychomotorische Fähigkeiten