About the position
Job description
Join TU Delft's MAVLab as a PhD candidate and develop safe learning methods that allow autonomous drones to improve their flight control through their own flight experience.
Job description
PhD on Safe Experience-Based Learning for Autonomous Drone Flight Control – TU Delft MAVLab
Conventional autopilots rely on fixed control laws and do not improve through accumulated flight experience. Human pilots, in contrast, become more capable as they gain experience. This project investigates how autonomous drones can similarly improve their flight control over time, based on their own flight experience, while remaining safe and reliable from day one.
The central research challenge is to learn and adapt flight-control laws without allowing the learning process to compromise the safety or performance of the drone. You will investigate approaches to safe learning and adaptation, including whether updates should take place during flight or between flights, and how the system can detect and handle potentially unsafe changes. The specific methods and safeguards remain open to research.
The research will be developed and evaluated on real flying drones (quad copters and VTOL winged drone, like the Rotating Wing Drone). The drone must be capable of autonomous flight from the outset, with its performance and learning progress assessed through real flight experiments. The project may also explore how experience-based learning can be extended to recognising and responding to dangerous situations.
You will join the Micro Air Vehicle Lab (MAVLab) in the Control and Simulation group of the Department of Control and Operations at TU Delft's Faculty of Aerospace Engineering. MAVLab has extensive experience in developing, building and testing autonomous aerial vehicles, including rotating-wing drones and projects such as NederDrone, which focuses on drone operations from ships. The lab combines fundamental research with a hands-on experimental approach, where new ideas are implemented and validated through real-world flight tests. You will work alongside researchers with complementary expertise in autonomous flight, vehicle design and control.
Job requirements
We are looking for a candidate with a strong scientific background and an interest in autonomous flight and learning-based control:
The following qualifications are required:
A master's degree in aerospace engineering, robotics, control engineering, or another relevant discipline.
A strong scientific background relevant to the project.
Strong programming skills, including proficiency in Python.
The ability and motivation to conduct independent scientific research, develop novel methods and publish scientific results.
An interest in autonomous systems and in improving their performance through learning and adaptation.
A willingness to implement and test your research on real flying drones and to learn how to organise and conduct experimental flight tests.
Good communication skills and the ability to collaborate effectively in an international, multidisciplinary research team.
Excellent written and spoken English.
The following are beneficial but not required:
Experience in control theory, control systems, or artificial intelligence (AI).
Experience in machine learning or reinforcement learning.
Knowledge of C or C++, or a willingness to learn these languages for embedded software development.
Practical technical skills, such as building, programming or integrating drones and onboard systems.
Experience with experimental flight testing or autonomous drone projects.
Familiarity with FPV drone flying.
The position is primarily focused on scientific research. Practical engineering and flying experience can help you organise flight tests and iterate more quickly, but prior drone-building or flying experience is not required. You should, however, be willing to learn the practical skills needed to demonstrate your research on real flying platforms.
TU Delft (Delft University of Technology)
This listing was collected from a public source and is reproduced here for
information only. Always confirm the details on the original posting before applying.
View the original posting