Projet de spécialité - Classification de signaux temporels pour le déclenchement d'airbags

De Ensiwiki
Révision de 18 novembre 2020 à 09:05 par Galiezc (discussion | contributions)

(diff) ← Version précédente | Voir la version courante (diff) | Version suivante → (diff)
Aller à : navigation, rechercher

Context: The creation of In&motion is based on a simple idea: embedded technology can revolutionize individual protectionsystems.Smart solutions, with increased ergonomics and adaptability, will signifi cantly improve our level of protection in theevent of a fall or an accident. Three passionate engineers—Valentin Honoré, Rémi Thomas and Pierre-François Tissot—graduating from ECAM,one of the most prestigious engineering schools in France, are at the origin of this company.Based in the heart of the French Alps in Annecy, In&motion is specialised in the development and industrialisationof smart protection solutions that can be adapted in the fields of sports, industry and health.In&motion is positioned as the leading partner of major brands who take advantage of our capacity to integrateour systems directly in their equipment. They can then focus on the promotion and marketing of the final product.The French startup has already launched 4 products:

* The smart ski airbag vest, approved by the International Ski Federation, and worn by more than one hundred athletes in Wolrd Cup and Europa Cup.
* In&motion provides also MotoGP riders with its airbag since 2016.
* An airbag system dedicated to regular motorcyclists (that got a CES Award 2017)
* A new generation mechanical airbag to equip horseriders via our partner Horse Pilot since 2017
* The 2020 Andalousia Rallye marks the launch of the new Off-road motorcycling system, adapted for the Dakar.


Goal: On se propose de réaliser de l'apprentissage automatique sur ces signaux temporels de manière à les classifier par type d'activité, afin d'améliorer la précision et la sensibilité du déclenchement.


Techniques involved: Pour cela nous utiliserons divreses techniques d'apprentissage automatique, comme par exemple : feature extraction, logistic regression, deep learning, clustering. Implémentation en python/R, keras/pyTorch.

Contact: Clovis Galiez clovis.galiez@grenoble-inp.fr