Journée Lab-STICC pôle IA & Océan

LabSTICC-IA-Ocean
Vendredi 18.03.2022
Horaires :
De 09:30 à 16:00

Adresse :

Pôle numérique Brest Iroise

Public cible

Etudiants
Doctorants
Enseignants chercheurs

Journée Lab-STICC pôle IA & Océan

Pôle numérique Brest Iroise, vendredi 18 mars


Programme :

  • 09h30 à 10h00 → café d’accueil
  • 10h00 à 10h30 → présentations du Lab-STICC (Christian Person) et du pôle “IA & Océan” (Pierre Tandeo)
  • 10h30 à 11h15 → présentation de l’équipe OSE (Abdesslam Benzinou) et d’un doctorant OSE (Amédée Roy)
  • 11h15 à 12h30 → discussion autour des posters
  • 12h30 à 14h00 → buffet
  • 14h00 à 14h45 → présentation de l’équipe M3 (Pierre Bosser) et d’un doctorant M3 (Julian Le Deunf)
  • 14h45 à 15h30 → présentation de l’équipe ROBEX (Luc Jaulin) et d’un doctorant ROBEX (Morgan Louédec)
  • 15h30 à 16h00 → réflexion autour de thèmes de recherche communs aux trois équipes
En savoir +

Présentations doctorant OSE :

  • Doctorant : Amédée Roy
  • Titre : Generative Adversarial Networks (GAN) for the simulation of seabird trajectories
  • Résumé :

Miniature electronic devices such as GPS have enabled ecologists to document relatively large amount of animal trajectories. Modeling such trajectories may attempt (1) to explain mechanisms underlying observed behaviors and (2) to elucidate ecological processes at the population scale by simulating multiple trajectories. Existing approaches to animal movement modeling mainly addressed the first objective and they are yet soon limited when used for simulation. Individual-based models based on ad-hoc formulation  end empirical parametrization lack of generability, while state-space models and stochastic differential equations models, based on rigorous statistical inference, consist in 1st order Markovian models calibrated at the local scale which can lead to overly simplistic description of trajectories We introduce a ‘state-of-the-art’ tool from artificial intelligence -
Generative Adversarial Networks (GAN) - for the simulation of animal trajectories. In this study, we aim on one hand to identify relevant deep networks architecture for
simulating central-place foraging trajectories and on the second hand to evaluate GAN benefits over classical methods such as state-switching Hidden Markov Models
(HMM).

Présentation doctorant M3 :

 

  • Doctorant : Julian Le Deunf
  • Titre : Traitement de données lidar bathymétriques avec des modèles d'apprentissage machine
  • Résumé :

Le traitement des nuages de points bathymétriques est une tâche critique pour l'élaboration des produits nautiques et son automatisation est un défi non résolu encore aujourd’hui ; l'objectif final étant de reporter les informations bathymétriques sur les cartes marines (possédant un statut légal). En particulier, le traitement des données topo-bathymétriques lidar est extrêmement complexe, et est encore aujourd'hui réalisé de manière totalement manuelle par les hydrographes et les opérateurs du Shom. Ces données sont également essentielles pour évaluer l'impact de l'érosion, le risque d'inondation et les effets du changement climatique sur le littoral. Elles fournissent des informations clés sur l'évolution des écosystèmes et des habitats sous-marins et sur l'impact des travaux de protection côtière passés ou prévus. Cette présentation montrera le modèle d'apprentissage machine mis en place pour essayer de résoudre ce problème ainsi qu'une expérimentation de la mise en place de ce processus sur de la donnée réel avec des opérateurs du Shom.


Présentation doctorant ROBEX :

 

  • Doctorant → Morgan Louédec
  • Titre → Interval Extended Kalman Filter: combining probabilistic and set membership filters
  • Résumé → The probabilistic and the set membership approaches in the localization problems are often seen as different and in competition. While in practice, the robots often rely on a probabilistic localization with a Kalman Filter for their online autonomous navigation, the set membership method is mainly used in post-process analysis. However, the Extended Kalman Filter (EKF), which has shown to be a precise method for nonlinear state estimation, may diverge if its linearization point is far from the true state. We propose a new filter, the Interval Extended Kalman Filter (IEKF) which benefits from the advantages of the EKF and the Interval Filter (IF). By linearizing the EKF in a feasible set generated by the IF, the IEKF is more likely to converge toward the state of the robot. This new filter has been applied for simulated underwater localization  scenarios.

Organisateur(s)

Équipe de Recherche :  OSE (Observations Signal & Environnement)

Département Mathematical and Electrical Engineering

Publié le 15.03.2022
 
 
 
 
 
1
 
2
 
3
 
4
 
5
 
6
 
7
 
8
 
9
 
10
 
11
 
12
 
13
 
14
 
15
 
16
 
17
 
18
 
19
 
20
 
21
 
22
 
23
 
24
 
25
 
26
 
27
 
28
 
29
 
30
 
31