[an error occurred while processing this directive]
Gallery
Vision







Visual Servoing




Amaury Dame

| Contact | Background and position | Research | Publications | Teaching | CV |

New contact

Amaury Dame is now "Research Assistant" in the Active Vision Group, Department of Engineering Science, University of Oxford

Email : adame@robots.ox.ac.uk
Web: Homepage

photo

Background and Position

In 2007, I received a master degree in industrial computer science and electrical engineering from INSA Rennes (INSA) and Research Master degree in signal and image processing (STI) from the University of Rennes 1.

In December 2010, I received the Ph.D degree in computer science from the University of Rennes after having work three years in the Lagadic research team in IRISA/INRIA Rennes Bretagne Atlantique under the supervision of Eric Marchand. The subject of my researches was to develop robust tracking and visual servoing algorithms.

I am currently searching for a postdoctoral position. My research interests include computer vision, active vision and visual servoing.

Research areas

In this thesis, we focused on both the use of optimization methods and robust alignement function to propose new algorithms for the tracking and the visual servoing problems. A metric derived from information theory, mutual information (MI), is principaly considered. Mutual information is widely used in multi-modal image registration (medical applications) since it is robust to changes in the lighting conditions and to a wide class of non-linear image transformation.

Mutual information based tracking

Principle

This work is mixing both information theory and differential tracking to propose a robust and accurate tracker. Mutual information has already been used in several tracking algorithm. In this work we propose a new approach that largely increases its robustness and accuracy. A new approach is also proposed to perform real-time tracking.

Validation

The validity of the approach has been validated through several experiments. The tracker has been evaluated on the metaio benchmark and gives very good results (details to appear). Many sequences have also been treated and shows the accuracy of the tracker as it is show in the video opposite. Since the estimation of the displacement is very accurate, the use of the proposed approach gives a new solution to augmented reality applications.

IEEE ISMAR'10 Best Paper Runner-Up Award
IEEE RFIA'10

Tracking through various appearance variations


download mp4 here

Mutlimodal
tracking


download mp4 here

Augmented reality
application


download mp4 here

Augmented reality
application


download mp4 here

Mutual information based visual servoing

Principle

We propose a new way to achieve visual servoing using directly the information (as defined by Shannon) contained in the images. In this work mutual-information is used as a new visual feature for visual servoing and allows us to build a new control law to control the 6 dof of the robot. Among various advantages, this approach does not require any matching nor tracking step, is robust to large illumination variation and allows to consider, within the same task, different image modalities.

Validation

Several platforms are available to perform the validation of our approaches. The MI-VS has been tested on a 6 dof gantry robot. The algorithm is converging with a 3D trajectory close to the geodesic and a final positioning error of about 0.1 mm in translation and 0.1 in rotation with a distance of 1 meter between the camera and the scene. Two typical experiments are presented opposite. The ICRA video that shows the robustness of the approach with respect to illumination variations and mutlimodality are available here.

IEEE ICRA'10
IEEE RFIA'10
IEEE ICRA'09

Mutual information based Visual Servoing


download mp4 here

Grasping using entropy based visual servoing


download mp4 here

Optimal detection and tracking of feature points using Mutual Information

Principle

We propose a new way to achieve feature points detection and tracking. An approach based on Mutual Information is propose to track each point with a strong robustness to illumination variations. Considering this tracking approach, an optimal detection of the points is proposed to choose the points that will be the more efficiently tracked.

Validation

The proposed approach has been tested on various image sequences and compared to the classical KLT approach. As the images opposite shows the proposed algorithm is far more robust to illumination variations.

IEEE ICIP'09

KLT

Proposed

Nominal conditions
photo photo
Global illumination variation
photo photo

Publications

Complete list (with postscript or pdf files if available)

Teaching

Computer vision courses at INSA (National Institute of Applied Sciences) Rennes.

Curriculum vitae

Click here to get a pdf version

[an error occurred while processing this directive]