Séminaire de TA Viet Cuong, doctorant de l'Institut MICA - Date : mercredi 22 novembre 2017, 09h00 - Lieu : "seminar room", Institut MICA, Hanoi University of Science and Technology

 

Intervenant :
M. TA Viet Cuong, doctorant en co-tutelle entre l'Institut MICA et l'équipe Pervasice Interaction de l'INRIA/LIG Grenoble

 

Date : mercredi 22 novembre 2017, 09h00
Lieu : salle "séminaire", 9ème étage, bâtiment B1, Institut MICA, Hanoi University of Science and Technology
Langue : le séminaire sera présenté en anglais

 

Résumé/Abstract:
With the popularity of smartphones and tablets in daily life, the task of finding user’s position through their phone gains much attention from both the research and industry communities. Technologies integrated in smartphones such as GPS, Wi-Fi, Bluetooth and camera are all capable for building a positioning system. Our works study the usage of Wi-Fi, motion sensors (inertial sensors) and Bluetooth for the indoor positioning task. With Wi-Fi, the context of training fingerprinting model on a few amount of data is explored. Several optimization layers, including alternating feature spaces, learning models and targets, are employed. The best results on a multi-floor indoor dataset are 93% floor accuracy and 5.12 mean distance errors. The results are comparable to related works, which are trained on large training datasets. In the second part, based on the Wi-Fi results, the outputs from Wi-Fi positioning system are combined with the inertial sensor tracking results. The work is aimed to fix the drifting errors from the inertial sensor tracking. Two combination approaches are proposed, which are Direct Adjust and Using Observation Model. In the third part, we study the capability of collaborative positioning by using both Wi-Fi and Bluetooth technology. The Bluetooth data is used to derived the user-to-user distance. The distance then is used to improve the Wi-Fi positioning output based on a Non-Temporal combination or a Temporal combination. The proposed approaches could reduce the position errors by 2m on a real world scenario.