Research group: Robotic Perception
My research work deals mainly with the direct exploitation of light intensities collected by a camera as input of control laws and object tracking algorithms. They are part of the research field of direct vision which aims to model and exploit the shortest possible link between the raw data acquired by the camera and the degrees of freedom to be controlled or the parameters to be estimated. The interest is also to get rid of the detection of features, their matching and the estimation of intermediate parameters.
I apply this general thread to unconventional vision (essentially unconventional field of view) for visual servoing, navigation and localization of various robots and, to a lesser extent, in virtual environments.
Beyond the fundamental aspects, I carry out this work in the main context of digital heritage, through the multidisciplinary research program E-Cathedrale around the digitization of monumental architectural heritage buildings (projects: FullScan, ASSIDUITAS, Transept, Athar3D) and the responsibility of the Technical Committee 19 "Computer Vision for Cultural Heritage Applications" of IAPR that I have been chairing since 2018 (co-organization of e-Heritage workshops and animation of the scientific community).
With more details, my main scientific contributions are distributed into three aspects: visual tracking, visual servoing and features matching and structure and/or motion estimation, of which details are also available from the side menu.
Here are the research projects in which I am or I was involved :
The thesis of my Habilitation to lead researches (HDR) makes a summary of my teaching and research activities as Associate Professor from 2011 to 2019. It proposes a survey on direct robot vision, focused but detailed under a unique formalism. It also gathers the main contributions I made with the PhD student I co-supervised. The summary of contributions is discussed and research lines are drawn from the short to the long terms.
Habilitation thesis (in French, December 2019) : PDF
My thesis proposes and studies formulations, adapted to omnidirectional vision, of features in problems of pose estimation, motion estimation and visual servoing.
PhD thesis document (november 2010): PDF (in French)