[an error occurred while processing this directive]
Andrew Comport is no longuer in the Lagadic project. After a postdoc at INRIA-Sophia Antipolis (ICARE), he is now a CNRS researcher at LASMEA Clermont Ferrand.
A. Comport. -
Towards a computer imagination: robust real-time 3D tracking of rigid and articulated objects for augmented reality and robotics . Thèse de l'Université de Rennes 1, mention informatique, Rennes, septembre 2005
The defence comitee was composed by Marie-Odile Berger (Chargé de Recherche, INRIA Loria), Pascal Fua (Professeur CVLAB, EPFL Lausane), Bradley Nelson (Professor of Robotics and Intelligent Systems à ETH-Zürich), Patrick Bouthemy (Directeur de Recherche, INRIA Rennes), Francois Chaumette (Directeur de Recherche INRIA, IRISA, Rennes), Éric Marchand (Chargé de Recherche INRIA, IRISA, Rennes)
In 1997, I graduated with Bachelor of Science (BSc) in Computer Science and in 2000 with a Bachelor of Engineering (BE) in Electrical and Computer Systems Engineering with Honours from Monash University Australia. In 2000, I undertook an exchange program with the Ecole Superieur de l'Electronic de l'Ouest (ESEO) in Angers, France. After graduation I worked as a Research Assistant for Ray Jarvis in the Intelligent Robotics Research Center(IRRC). In October 2001 I started a PhD in the Lagadic research project at IRISA/INRIA Rennes. My supervisors are François Chaumette and Eric Marchand.
The topic of my thesis is "Virtual Visual Servoing for Augmented Reality". Initially this thesis has been based on using traditional visual servoing techniques to calculate the pose of a camera. More precisely it has involved localizing the viewpoint between a camera and a model of the observed scene by using the image acquired by the camera. This research has also taken into account different types of visual information such as distances to lines, ellipses, cylinders and spheres or any combination of these. Further to this a visual servoing control law has been developed which is statistically robust with respect to noisy measures and potential occlusions. Current research has involved considering more complicated models such as articulated objects by relaxing rigidity constraints. Spatio-temporal object representations are also being considered. These methods are directly related to the research areas of 3D Tracking, Visual Servoing and Augmented Reality and have many applications in diverse domains such as production, medicine and entertainment industries.[an error occurred while processing this directive]