Project

Discipline

wind; turbulence; certification; WindShape; drone; weather

Lead
Nous proposons de mesurer des écoulements météorologiques en milieu urbain (EPFL) et d'utiliser des outils de Machine Learning (ETHZ) afin d'entraîner un instrument d'invention Suisse, une soufflerie pixelisée (WindShaper), à reproduire fidèlement ces conditions. L'objectif est de développer des environnements d'essais et de certification pour les drones du monde entier (HES-SO).
Lay summary
Les drones rempliront bientôt notre écosystème aérien dans les domaines de l'imagerie/cartographie, de la livraison de colis et du transport de passagers. Les drones devront fonctionner dans des conditions atmosphériques arbitraires, en particulier dans des conditions météorologiques défavorables lors de situations d'urgence. Dans les villes, ils devront surmonter les turbulences urbaines ainsi que les turbulences des véhicules routiers. Nous avons soumis une proposition à l'Office Fédéral de l'Aviation Civile (OFAC) de développer des protocoles de certification de drones dans un environnement contrôlé, en utilisant un Simulator of Autonomous Flight Environments (SAFE) basé sur une soufflerie pixelisée (WindShaper), inventée en Suisse et commercialisée par une entreprise genevoise, WindShape. SAFE soumettra les drones à des vents d'intensité et de direction variables qui reflètent des situations réelles. Pour atteindre ces objectifs, SAFE doit reproduire les conditions environnementales réelles. Dans ce contexte, deux objectifs clés seront poursuivis dans le contexte de ce projet: 1. fournir à cette installation des données réelles de turbulence atmosphérique à l'échelle des drones, 2. développer une méthodologie basée sur du Machine Learning pour reproduire ces flux dans notre soufflerie pixelisée. SAFE sera la première installation au monde à garantir la navigabilité des drones en fonction de leur capacité à rester en altitude dans des conditions de vent et de météo données. L'infrastructure (matériel), les flux (logiciels), ainsi que la procédure de test et de certification (processus) pourront ensuite être commercialisés dans le monde entier, soit en tant que service, soit en tant que franchise.

Direct link to Lay Summary

Last update: 14.05.2020

FNS Bridge SAFE 2020-2021

Author	Noca, Flavio
Persistent Identifier (PID)	10.26037/yareta:d5aslhhavvg2tgxzx6x2r6dkxy
Repository	CHACHACHA
Abstract

1. Master theses- David Sollander (EPFL/HES-SO)- Daniil Morzhakov (EPFL/Voliro)- Marco Camozzi (HES-SO)- Geshanth Visvaratnam (HES-SO)- Aurélien Walpen (HES-SO)2. Preliminary output data from HES-SO

Active participation

Active participation

Communication	Title	Media	Place	Year

Drones will soon fill our aerial ecosystem in the field imaging/cartography, parcel delivery (La Poste, Amazon), and passenger transport (Uber Elevate). Drones will need to operate around the clock in arbitrary atmospheric conditions, especially in adverse weather conditions during emergency situations. In cities, they will be required to overcome urban canyon as well as road vehicle turbulence. At higher altitudes, winds can be unpredictable: drones are much smaller than conventional aircraft and are thus more sensitive to weather conditions.Today, traditional drone testing techniques are of poor quality and do not reflect conditions that may be encountered in real world applications. Test drones are either flown outdoors in not well documented, uncontrolled and unpredictable weather conditions (and quite remote from the observer), or tightly strapped onto a support in a conventional wind tunnel with laminar and uniform wind flows. Such wind flows have been devised for conventional aircraft and they are inadequate representations of atmospheric conditions relevant to drones. In summary, existing tests entail a high risk of reaching false conclusions about drone performance and introduce additional risks on drone design that may lead to catastrophic situations.We have submitted a proposal to the Federal Office of Civil Aviation (FOCA) to develop drone certification protocols in a controlled environment, using a Simulator of Autonomous Flight Environments (SAFE) based on a multifan wind facility commercialized by a Swiss company, WindShape. The wind facility consists of an array of a large number of fans that may be arranged in various patterns on demand. SAFE will subject drones to winds of variable intensity and direction (as well as various weather conditions such as rain, snow, hail, fog etc.) that reflect real world situations. These tests would then rate drones according to their capacity in maintaining a proper flight attitude and tackling flight perturbations in an urban, countryside, or high altitude environment.In order to achieve these goals, SAFE needs to replicate real environmental conditions. In this context, two key steps need to be undertaken: 1. supply this facility with actual atmospheric turbulence data at drone scales (in particular, those encountered in an urban environment), and 2. develop a methodology to reproduce these flows in a multifan facility. The objective of this Bridge proposal is to achieve these two steps by : - Collecting actual urban turbulence data using a facility (motus.epfl.ch) installed at the Ecole Polytechnique Fédérale de Lausanne: a meteorological mast, instrumented with sonic anemometers with high sampling frequency, which provides wind intensity and direction in an urban setup; - Training thousands of fans in a WindShape facility to produce given 3D flows in space and time, using Machine Learning algorithms and a vast experimental data set collected within the facility.Ultimately, the facility will be able to generate realistic atmospheric flow fields and in particular, genuine urban wind turbulence. SAFE will be the first ever facility in the world to ensure drones airworthiness according to their capacity in staying aloft in given wind and weather conditions. The infrastructure (hardware), the flowfields (software), as well as the testing and certification procedure (processes) can then be commercialized worldwide, either as a service or as a franchise.