Fusion of Audio and Temporal Multimodal Data by Spreading Activation for Dweller Localisation in a Smart Home

Pedro Chahuara (1) , François Portet (1) , Michel Vacher (1)

1 GETALP - Groupe d’Étude en Traduction Automatique/Traitement Automatisé des Langues et de la Parole

In this paper, an approach to locate a person using non visual sensors in a smart home is presented. The information extracted from these sensors gives uncertain evidence about the location of a person. To improve robustness of location, audio informa- tion (used for voice command) is fused with classi- cal domotic sensor data using a two-level dynamic network and using an adapted spreading activation method that considers the temporal dimension to deal with evidence that expire. The automatic lo- cation was tested within two different smart homes using data from experiments involving 25 partici- pants. The preliminary results show that an accu- racy of 90% can be reached using several uncertain sources.

Domaines: Intelligence artificielle [cs.AI]

Pedro Chahuara, François Portet, Michel Vacher. Fusion of Audio and Temporal Multimodal Data by Spreading Activation for Dweller Localisation in a Smart Home. STAMI Series, Space, Time and Ambient Intelligence, Jul 2011, Barcelona, Spain. pp.17-22. ⟨hal-00953558⟩