Team: Rainbow IRISA/Inria Rennes
Team website: team.inria.fr/rainbow
How to apply:
Interested candidates are requested to apply via this form. The position will remain open until a suitable candidate is found.
This thesis is conducted in collaboration with Airbus Defence and Space as part of the Lichie project funded by BPI. It aims at defining methods to control the motions of one or several satellites according to the content of the images acquired by their imaging system. The methods selected will be based on visual servoing techniques.
Visual servoing consists of controlling the movements of a dynamic system using visual information acquired by a vision system and integrated within closed-loop control laws. This classically poses various problems: which visual information to retain depending on the task to be performed and the constraints of the robot (number of degrees of freedom to be controlled, dynamics, trajectory constraints, etc.), how to relate this visual information to the measurements made in the images, how to extract these measurements from the images, and finally how to elaborate an appropriate control law.
Within the scope of this thesis, one or multiple sensing devices (cameras) are embarked on each satellite whose orientation can be controlled (3 degrees of freedom) in order to focus on an object of interest, under the constraint of a fast scrolling movement (several km/s). In the context of a constellation of observation satellites, these satellites will have to cooperate in order to focus their camera on the object of interest in an optimal way (oblique versus azimuthal shots for example). A collaborative estimation of the object’s position or trajectory can then be envisioned.
Several applications are envisaged: tracking of static or mobile objects on the ground or in flight, observation, tracking and localisation of other (possibly non-cooperative) satellites.
The methodological developments will be carried out in C++ using the ViSP software library and will be validated in simulation and then via experiments on real images and models.
Master’s or engineering degree in computer science, computer vision, robotics, applied mathematics, aeronautics (or related fields)
- knowledge in computer vision, robotics and automation
- knowledge of astronautics would be a welcome addition.
- mathematical skills, optimization
- C++ programming
- willingness for experimentation
The position is full-time for 3 years and will be paid according to French salary rules for PhD students.