Skip to Main content Skip to Navigation
Theses

Schémas de vol adaptatifs pour l'exploration de nuages par une flotte de drones : principe, mise en œuvre et expérimentations

Abstract : Atmospheric scientists are constantly seeking to acquire new data that can improve their models of atmospheric phenomena, especially clouds. Current methods are insufficient to collect adequate measurements of cloud dynamics and microphysical parameters related to cloud formation, generating large uncertainties in model formulation. This lack of in-situ data leads meteorologists to find new collection methods.The use of UAVs is now widespread and many applications are emerging in different contexts. Although the use of a single UAV is very popular, the deployment of UAV fleets is still uncommon and is mainly limited to exploring and mapping unknown static environments. A fleet could find its usefulness in other more complex applications such as the monitoring of dynamic phenomena (e.g. oil puddles on the sea, plumes of smoke from a factory, atmospheric phenomena). Research on the coordination of a UAV fleet and the exploration of dynamic environments is not yet complete and many contributions can be made to the current problems in this field.The objective of this thesis is to provide solutions and strategies to explore a dynamic environment such as the evolution of a cloud with a fleet of UAVs, thus providing better temporal and spatial coverage than with a single UAV. This calls for the development of a UAV control and planning architecture that ensures the cooperation of UAVs to carry out the mission in the best possible way. The constraints associated with this type of environment and mission limit the collective work of the fleet. For robustness and efficiency reasons, the system's mechanisms are implemented in a distributed manner, where the UAVs embark the planning processes and communicate directly with each other, rather than through a single station.This thesis was carried out in close collaboration with the NEPHELAE project which aims to collect data in cumulus clouds in order to reconstruct a 4D spatio-temporal model of its evolution. Knowing that classical flight patterns used in autopilots are not efficient to explore such dynamic environments, the main contribution of this thesis is the development and implementation in the PAPARAZZI system of adaptive flight patterns. These flight patterns use real-time sensor measurements to adapt the UAV trajectories to the cloud to be mapped. This action is performed onboard the UAV and without the intervention of an operator. The drone's behavior changes according to the pattern used, enabling the tracking of the cloud edge, the construction of a dense 3D map or the determination of the cloud core.The validation of these new navigation functions was carried out through different simulations combining UAVs simulated in a static then dynamic cloud environment. Subsequently, a first hybrid experiment was carried out before deploying the fleet during a measurement campaign in Barbados in early 2020. This campaign enabled a large number of exploratory flights and cloud tracking in real conditions. In addition to providing results and suggestions for improvements in adaptive flight patterns, it allowed atmospheric scientists to collect important data on clouds that had not been observed until today. In particular, this experiment made it possible to follow a cloud edge with several UAVs simultaneously, thus achieving a first in terms of data collection in a cloud.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-03172221
Contributor : Abes Star :  Contact
Submitted on : Wednesday, March 17, 2021 - 3:38:17 PM
Last modification on : Thursday, March 18, 2021 - 3:05:53 AM

File

2020VerduTitouan.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03172221, version 1

Collections

Citation

Titouan Verdu. Schémas de vol adaptatifs pour l'exploration de nuages par une flotte de drones : principe, mise en œuvre et expérimentations. Automatique / Robotique. INSA de Toulouse, 2020. Français. ⟨NNT : 2020ISAT0010⟩. ⟨tel-03172221⟩

Share

Metrics

Record views

76

Files downloads

26