The typical duties of micro-drones can be quite diverse: identification of ground targets, search and localization of targets, drop of payloads, etc. A mission for one or several drones is made of a set of such tasks. The associated mission planning problem for a patrol of micro-drones consists in allocating these tasks to the drones and scheduling their execution while minimizing the makespan of the mission. The execution of the tasks may fail and they must be planned again in such a case; this fact introduces uncertainties during the mission which cannot be taken into account before its realization.
This thesis proposes to use a full centralized mission planning system for micro-drones. The problem is modelled by a constraint satisfaction problem and is solved with complete methods able to offer proofs of optimality of the solutions and heuristic methods that can quickly obtain good solutions on bigger instances. The management of uncertainties is handled by a reactive architecture combined with an online planning module. The complete planning system has been integrated in the Paparazzi system which can actually control a set of drones, by simulation or in real conditions. The presented algorithms are experimented on various scenarios and missions with promising results.
- Plannification de mission pour une patrouille de micro-drones. Journées Francophones Planification Décision Apprentissage. June 2009
- Automated Mission Planning for a Swarm of MAV. 4th US-European Competition and Workshop on Micro Air Vehicle June 2009 (Accepted)
- Automated Planning for a Fleet of Micro Air Vehicles. 3rd US-European Competition and Workshop on Micro Air Vehicle System & European Micro Air Vehicle Conference and Flight Competition. September 2007
- Safety Memo. 3rd US-European Competition and Workshop on Micro Air Vehicle System & European Micro Air Vehicle Conference and Flight Competition. September 2007. (pdf file)