Thesis (m/f/d): Learning-Based Control Allocation For Morphing UAVs

German Aerospace Center (DLR); Cochstedt; On-site; EUR 8,900 - 12,000; Full time

Job summary

The Institute of Flight Systems at the German Aerospace Center (DLR) invites a Master’s student to work on a thesis developing a safe, learning-based control-allocation approach for a trailing-edge morphing-wing UAV under model uncertainty. The project focuses on meeting primary control objectives while exploiting actuator redundancy and incorporating safety mechanisms.

The work is jointly supervised by DLR and Prof.

Qualifications

• Enrolled in a Master’s program in Computer Science, Control Engineering, Robotics, Aerospace Engineering or a related field.
• Hands-on experience with reinforcement learning is mandatory, ideally with policy-gradient methods such as PPO or SAC.
• Practical experience with Gymnasium or comparable RL environments is mandatory.
• Good understanding of control engineering related to flight control design is a plus.
• Strong Python programming skills.

Responsibilities

• Review learning-based methods for safe and feasible control allocation under model uncertainty.
• Compare uncertainty-handling approaches (adaptive, observer-based, set-based, statistical learning) by guarantees, assumptions and real-time capability.
• Develop a learning-based control-allocation concept for a trailing-edge morphing-wing UAV under model uncertainty.
• Integrate safety mechanisms (e.g., CBF-based safety filters) and strategies to keep the allocation feasible (backup strategies, slack variables, feasibility certificates).
• Train and benchmark the developed approaches under varying operating conditions and model uncertainties.
• Analyze stability, safety and allocation feasibility.

Skills

Reinforcement learning

Policy-gradient methods (PPO/SAC)

Python programming

Flight control knowledge

Independent, structured working

Education

Master's program in CS/Control/Aerospace

Tools

Gymnasium

Job description

Welcome to the Institute of Flight Systems. Our work focuses on the interaction between aircraft configuration, pilots and modern flight system technology. From flight dynamics to unmanned aerial vehicles, from simulation to real flight tests - we analyse, test and develop innovations that will shape the flying of the future.

What To Expect

This thesis develops a safe, learning-based control-allocation approach for a trailing-edge morphing-wing UAV under model uncertainty. The allocation must meet the primary control objectives while exploiting actuator redundancy for secondary goals. A central part is handling model uncertainty, e.g., via adaptive methods, disturbance observers, set-based methods or statistical learning. Another focus is allocation feasibility, covering backup strategies, slack variables, adaptive weighting, parameter optimization and explicit feasibility certificates. The thesis is jointly supervised by DLR and Prof. Mayank Shekhar Jha (CRAN, CNRS, Université de Lorraine, France).

Your tasks

• Review learning-based methods for safe and feasible control allocation under model uncertainty
• Compare uncertainty-handling approaches (adaptive, observer-based, set-based, statistical learning) by their guarantees, assumptions and real-time capability
• Develop a learning-based control-allocation concept for a trailing-edge morphing-wing UAV under model uncertainty
• Integrate safety mechanisms (e.g., CBF-based safety filters) and strategies to keep the allocation feasible (backup strategies, slack variables, feasibility certificates)
• Train and benchmark the developed approaches under varying operating conditions and model uncertainties
• Analyze stability, safety and allocation feasibility

Your profile

• Enrolled in a Master’s program in Computer Science, Control Engineering, Robotics, Aerospace Engineering or a related field
• Hands-on experience with reinforcement learning is mandatory, ideally with policy-gradient methods such as PPO or SAC
• Practical experience with Gymnasium or comparable RL environments is mandatory
• Good understanding of control engineering specifically related to flight control design is a plus
• Strong Python programming skills
• Independent, structured way of working and a strong interest in combining RL, control and safety-critical autonomous systems

We look forward to getting to know you!

If you have any questions about this position (Vacancy-ID 6793) please contact:

Mark Spiller

Tel.: [phone]