Ambiant intelligence in the habitat

Publié le ven 9 Oct 2020

French version

Project goal: Measurement of human activities in the habitat

Low-cost, non-intrusive, non-binding IoT solutions

KeywordsApplication fieldsMethodologyKnow-howScientific issuesResultsPartnershipsContacts

Keywords :

Human activity measurement, IoT-Internet of Things, information fusion, false positive management, uncertainty management

Application fields

  • Home care for people in vulnerable situations

    • home measurement of modalities related to the inhabitant fragility
    • Referal system for
      • the person itself
      • the family
      • caregivers
      • medical and welfare professionals
  • Technology parts :
    • Indoor location by non-supported and non-intrusive devices
    • Actimetry
    • Measurement of habits and change of habits

Methodology

  • Integration of expert knowledge
  • Multi-modal, multi-source information fusion by IoT
  • Management of imperfect information: imprecise, uncertain, incomplete, contradictory
  • Decision support integrating the development of dashboards adapted to each user profile
  • Experimental validation
  • Privacy and health data regulation where applicable

Know-how involved in the project

  • Techniques for the identification of situations based on habitat measurements :
    • Information fusion and associated theories : possibility theory, probability theory, evidence theory .
    • Learning, representation, and use of high-level information
    • Uncertainty management, data incompleteness
  • Mastery of the main Machine Learning techniques of Artificial Intelligence and ability to select the one adapted to the application.
    • deep learning
    • data mining
    • supervised and unsupervised classification (FCM, KNN, Random Forest, SVM)
  • Rapid prototyping of sensitive objects embedding sensors and communicating objects. Mastery of embedded tools like nanocomputers and microcontrollers.
  • Rapid prototyping of applications for connected and intelligent objects.

Scientific problematics

  • Measurement of human activity (actimetry) from sensors preferably non-supported and non-intrusive.
  • Modeling and identification of human situations and activities, management of related imperfect information
  • Explicability : how to bring semantics and meaning to the measurement.

Results

This project resulted in the production of an abstract habitat representation model that allows automatic identification of situations. This approach is being applied in the creation of an intelligent showroom.

Publications :

An approach to express measurement results in ontologies of facts
Eric Benoit,
Cedric Deffo SikounmoStéphane Perrin.
IMEKO world congress, Sep 2018, Belfast, United Kingdom, ⟨10.1088/1742-6596/1065/7/072007⟩

Uncertainty management of situations in a housing use context
Cedric Deffo SikounmoStéphane PerrinEric Benoit
2017 IEEE Int. Conf. on Computational Intelligence and Virtual Environments for Measurement Systems and Applications (CIVEMSA 2017), Jun 2017, ANNECY, France. pp.129-134, (10.1109/CIVEMSA.2017.7995314)

Modeling of the situations in an intelligent connected furnitures environment
Cedric Deffo SikounmoEric BenoitStéphane Perrin.

The Tenth Int. and Interdisciplinary Conf. on Modeling and Using Context (CONTEXT 2017), Jun 2017, PARIS, France. pp.381-394, ⟨10.1007/978-3-319-57837-8_32⟩

States measurement in a context of intelligent connected furnitures
Cedric Deffo SikounmoEric BenoitStéphane Perrin.

2016 Joint IMEKO TC1-TC7-TC13 Symposium « Metrology across the Sciences: Wishful Thinking? », Aug 2016, Berkeley, USA, ⟨10.1088/1742-6596/772/1/012003⟩

Partnerships

  • Haute-Savoie Departmental Council (CD74)
  • Savoie Mont Blanc council
  • Annecy Genevois Hospital Centre (CHANGE)
  • Industrial partners : Miliboo, Somfy

You wish to meet us and/or to study an R&D partnership :

Contacts :