Thèse de doctorat:

Un de nos consultants, en l’occurence Almoctar Hassoumi vient de soutenir sa thèse de doctorat ce Lundi 25 Novembre 2019.


 Eye-tracking tools estimate the locations in a scene where a user is fixating on. They are used in various domains including human-computer interaction (HCI) and learning transfer. As an example, gaze-based text entry allows interacting with computing
systems remotely without touching the interface. They are also used to comprehend the visual behaviors of a pilot searching for information in a cockpit. However, a number of barriers still exists and makes these devices less accurate and difficult to use in daily activities. One of these problems is the shift between the actual and the estimated position of the user’s point-of-regard, which systematically comes from the eye-tracking systems’ accuracy. Following recent advances, there is an increasing interest in affordable systems that have the potential to be more accurate and, researchers are continually investigating novel approaches.
This thesis covers different issues of eye movement research. It proposes the use of novel approaches as a step towards overcoming these accuracy issues. More specifically, we introduce novel strategies for detecting mapping functions for gaze estimation and calibration-free gaze interaction. In addition to proposing frameworks and strategies for improving accuracy, new calibration procedures and patterns are also revealed and discussed. In this thesis, we address these issues in three different ways: calibration and mapping functions, Human-computer Interaction using the eyes, visualization and exploration. We present four main contributions. First, we present a new method for calibrating state-of-the-art eye trackers with better accuracy. Second, we present a new gaze-based authentication method which works without any prior calibration, and can be extended to any alphanumeric-based input modality. Third, we present an uncertainty visualization approach. Finally, a method of analyzing eye
movements data and aircraft trajectories using a novel brushing technique is proposed.

DEvant le jury composé de :

Olivier Le Meur

Maitre de conference HDR Univ Rennes CNRS IRISA - Rapporteur

Romain Vuillemot

Maitre de conference - Ecole Centrale de Lyon - Jury

Mickael Causse

Professeur - ISAE – Jury

Vsevolod Peysakhovich

Enseignant Chercheur - ISAE - Co-directeur de these

Enkelejda Kasneci

Professeur University of Tübingen - Rapporteur

Jean-Pierre Jessel

Professeur - IRT - Président

Jean Michel Boucheix

professeur, université de Bourgogne – Jury

Christophe Hurter

ENAC | Microsoft - Directeur de thèse