[Opt-net] PhD position in space relative dynamics and model predictive control techniques, Toulouse, France

[Jean-Claude Berges]

A PhD position will be available from (earliest) September 2010 at the 
French space agency (CNES) in Toulouse, France. The home Laboratory will 
be the LAAS-CNRS ( Laboratoire d’Analyse et d’Architecture des Systèmes 
Toulouse MAC (Methods and Algorithms in Control) Group: 
http://www.laas.fr/laas09/MAC-EN/ ). The duration of this contract is 
one year, renewable twice. The candidates will be funded by the CNES 
(French space agency) with a co-financing agreement with EADS/Astrium.
You can read more about the position at these links:
http://www.cnes.fr/web/CNES-en/7430-research-grants.php
http://www.cnes.fr/web/CNES-en/7441-how-to-apply.php
and about the subject description at these links:
http://www.cnes.fr/web/CNES-en/7432-topics-2009.php?view=item&item=2285
Subject short description :
The candidate will contribute to the development of autonomous orbital 
rendezvous guidance algorithms in the general framework of the model 
predictive control techniques. To provide rendezvous guidance algorithms 
with robustness property, an intuitive idea frequently used in the space 
industry consists in the following iterative procedure:
1. Generating a constrained maneuvers plan that leads the deputy vehicle 
to reach a final relative state in fixed time from its current state at 
the time t0.
2. Performing the first maneuver of the plan at the time t1 > t0.
3. Estimating the new current relative state at the time t2 > t1 (e.g. 
just after performing the first maneuver).
4. From this new state, generating a new plan of maneuvers to reach the 
target, and return to Step 2.

Although this heuristic method looks like a closed loop strategy, it is 
not based upon a formal and rigorous analysis that solely ensures 
stability and robustness with respect to uncertainties and allows 
performances quantification.
To cope with these issues, we intend to tackle the guidance problem in 
the general framework of the model predictive control (MPC) techniques. 
Albeit these methods take up the reasoning described above e.g. 
optimizing control actuation under constraints on a receding horizon 
they are mainly designed to take advantage from the feedback pieces of 
information about the gap between nominal and actual positions and also 
from the control sequence itself so that it becomes possible to online 
adapt the control law, soften the constraints or even evolve the model 
used in the algorithm. It is noteworthy that the feedback concept used 
in this framework is set upon a specific control structure, adaptive or 
not, for which the cost function which quantifies the rendezvous 
accuracy will be an influent factor on the stability of the whole 
system. The prospective nature of this way of investigation consists in 
replacing the previous heuristics by a formal approach that sets the 
stability, ensures robustness against uncertainty and provides good 
online performances.
Requirements:
For thesis contracts, university candidates or graduates of a Grande 
Ecole or engineering school must hold a Master 2 degree awarded by a 
university, or a suitable degree enabling them to enrol for their thesis 
at the relevant doctoral school. To this subject, knowledge on space 
relative dynamics and control techniques will also be appreciated.
Application:
The application deadline is 31th of March 2010 . The successful 
candidates might be selected for an interview process in spring and have 
to be available for a start between September and December 2010.
Further enquiries should also be directed to these email address: 
jean-claude.berges at cnes.fr, alain.theron at laas.fr, 
christophe.louembet at laas.fr .
BERGES Jean-Claude
CNES
DCT/SB/MO
18 avenue Edouard Belin
31401 Toulouse Cedex 9, France