Interested in working with us?
We are always looking for motivated students and researchers to join our group. Please follow carefully the instructions below when applying for a postion, otherwise you may not receive a reply, sorry for that!
We help each other to thrive. We believe that everybody has a unique talent and we help them find their talent and enjoy doing research with us. Mental health of the group members is our number-one priority.
We would love to have new members from different backgrounds and from minority groups. Our vision is to have a team that is as diverse as possible. We are commited to provide a safe environment for everyone to enjoy doing research with us.
BSc and MSc students (TUM students)
If you are doing your BSc or MSc studies at TUM and are interested to do your thesis with us, fill in this form with your CV and a 1-page proposal about the topic of your interest. Please specify in the proposal, how the proposed topic is related to the research of our group.
Project Description
This project focuses on the development and validation of a reinforcement learning based locomotion controller for a tendon driven underactuated biped, implemented in the Nvidia Isaac Sim environment. The primary objective is to achieve stable and energy efficient walking and running gaits, leveraging the compliant dynamics of the tendon system. Students will construct a physically realistic simulation of the robot, modeling the compliant tendon actuation either as chains of rigid bodies or using other appropriate approximations supported by Isaac Sim. The robot model will be implemented as a USD file compatible with Isaac Sim, with integrated sensor models and actuation interfaces. Using Isaac Lab, students will train a reinforcement learning policy that stabilizes bipedal locomotion while minimizing long-term energy expenditure. Emphasis will be placed on developing reward functions that promote both gait stability and energy efficiency, reflecting the advantages of the underactuated, compliant leg design. A key part of the project will involve tracking mechanical energy flows within the simulation to validate energy reuse through passive dynamics. Students will perform quantitative analysis comparing energy consumption under different gait regimes and control policies. Deliverables will include a functional Isaac Sim model of the biped, training results from Isaac Lab, visual demonstrations of the learned gaits, and documented energy calculations. The project offers experience in simulation, reinforcement learning, compliant system modeling, and quantitative analysis of locomotion efficiency.
For more information about this project in collaboration with RoboTUM, visit this link
Prerequisites
- Master-level studies in Electrical Engineering, Informatics, Computer Science or any relevant program
- Good knowledge of robotics software development, especially dynamics, motion planning and control.
- Practical experience with ROS
- Preferable to have prior experience with Reinforcement Learning
- Excellent C++/Python programming skills
- Ability to work well structured and organized
Workplace
Georg-Brauchle-Ring 60-62, 80992 Munich
Contact
Shafeef Omar (shafeef.omar(at)tum.de)
Project Description
This project addresses the challenge of transferring locomotion policies from simulation to real hardware using the TRON 1 robotic system as the primary validation platform. Building on an existing training pipeline developed in the DODO Lab in collaboration with candidate PhD researcher Vasilije Rakcevic, students will investigate and implement state-of-the-art Sim2Real techniques to bridge the gap between idealized simulation and physical deployment. The first phase involves a comprehensive review of Sim2Real challenges, including domain shift, sensor and actuator noise, contact model inaccuracies, and overfitting to simulation artifacts. Students will study and apply techniques such as domain randomization, curriculum learning, encoder regularization, adversarial domain adaptation, and hybrid fine-tuning strategies. The goal is to systematically improve the robustness and generalization of learned control policies. Using TRON 1 hardware, students will develop real-world perturbation profiles and design transfer metrics to quantify trajectory fidelity, control divergence, and task success rates. Ablation studies will evaluate the effectiveness of each technique. Sensor and actuator interfaces will be tested for timing and fidelity to ensure compatibility with the simulation environment. Once the methods are validated on TRON 1, the final phase will transfer the best-performing pipeline to the Forrest robot platform. Deliverables will include a modular toolkit for Sim2Real training and transfer, full documentation, and robust locomotion policies deployable across both systems. The project offers hands-on experience in modern reinforcement learning, robotic simulation, and real-world control validation.
For more information about this project in collaboration with RoboTUM, visit this link
Prerequisites
- Master-level studies in Electrical Engineering, Informatics, Computer Science or any relevant program
- Good knowledge of robotics software development, especially dynamics, motion planning and control.
- Practical experience with ROS
- Preferable to have prior experience with Reinforcement Learning
- Excellent C++/Python programming skills
- Ability to work well structured and organized
Workplace
Georg-Brauchle-Ring 60-62, 80992 Munich
Contact
Shafeef Omar (shafeef.omar(at)tum.de)
Internship for external students
We would be happy to discuss the potential of having you in our team for your BSc or MSc thesis. If you are a student in Germany, we may be able to give you a working student (HiWi) contract, if our funding situation allows. However, if you are a student outside of Germany, we can only host you if you are able to secure funding for your stay from your host university or a third party. In these cases, please fill in this form with your CV and a 1-page proposal about the topic of your interest . Please specify in the proposal, how the proposed topic is related to the research of our group.
PhD candidates
We are always interested in hiring new motivated PhD candidates, as long as our funding situation allows. If you already have a master degree (or graduating soon) and are interested to do your PhD with us, please fill in this form with your CV and a 1-page proposal about the topic of your interest. Please specify in the proposal, how the proposed topic is related to the research of our group.
Note: At the moment, we do not have fundung available for hiring new PhDs.
Postdocs
We are always interested in hiring new motivated Postdocs, as long as our funding situation allows. If you already have a PhD degree (or graduating soon) and are interested to continue in academia with us, please fill this form with your CV and a 1-page proposal about the topic of your interest. Please specify in the proposal, how the proposed topic is related to the research of our group.
Note: At the moment, we do not have funding available for hiring new Postdocs. If you are currently outside of Germany, you can apply for the TUM Global Postdoc Fellowship.