Students Projects
Master Thesis / Bachelor Thesis / Internship / Forschungspraxis (EI)
Important! When applying to any project, please always include your CV and academic transcript to allow us to evaluate your application.
Master’s Thesis: Unsupervised Deep Learning Approaches for Generalizable Myoelectric Control
Current myoelectric control strategies often require extensive supervised training data and show limited generalization across users. In this Master’s thesis, you will explore unsupervised deep learning approaches to improve gesture estimation from muscle activity without relying heavily on labeled data. Using surface electromyography (sEMG) recordings, you will develop and evaluate models capable of learning robust, transferable representations of muscle activation patterns. The ultimate goal is to enable more adaptive and user-independent control of myoelectric interfaces.
Your Tasks
- Review the state of the art in self-supervised and unsupervised learning for biosignal decoding
- Develop deep learning architectures for feature extraction from unlabeled sEMG data
- Integrate learned representations into gesture classification or regression frameworks
- Evaluate model generalization across participants and recording sessions
Your Qualifications
- Proficiency in Python
- Strong background in deep learning (preferably experience with LLMs, transformers, or time-series modeling)
- Preferably, knowledge in the processing of sEMG signals
- Strong interest in neural interfaces, representation learning, and human-machine interaction
Master’s Thesis: Object Detection Through Proprioception of Adaptive Synergistic Grasps in a Robotic Hand
Synergistic grasping is a neural concept that balances forces during human grasping. The same principle can be applied to highly underactuated robotic hands, in which the fingers adapt passively to the shape of grasped objects and distribute forces in a balanced manner through the hand's mechanical design. The goal of this project is to explore the extent to which the joint angles of an adaptive hand can yield information about the object that has been grasped.
Your Tasks
- Literature review on proprioception for robotic hands
- Setting up a multi-body simulation of a synergistic robotic hand
- Evaluating grasp configurations for the YBC grasp dataset
- Training a classifier/regressor to distinguish objects from the benchmark dataset
Your Qualifications
- Background in Mechanical Engineering, Robotics, Mechatronics, or Biomedical Engineering
- Strong coding skills in Python
- Previous experience with MuJoCo or other multi-body simulators is beneficial
Open applications
We welcome highly motivated students who wish to contribute but find that their interests are not reflected in our current projects.
If you have an idea in mind, please reach out to us with your proposal. Always include your CV and academic transcripts to allow us to evaluate your application. [Contact us]