IROS 2004 Tutorial on Advanced Visual Servoing

NEW: Corrected and updated version of the tutorial notes and videos.

Tuesday, September 28: 13:45-18:15 (half-day)
Meeting Room 4, 2nd floor, Sendai International Center
Organizer(s):The EURON Interest Group on Visual Servoing
Contact:
- Philippe Martinet, Univ. Blaise Pascal, Clermont-Ferrand (France)
- Enric Cervera, Jaume-I University of Castelló (Spain)

Lecturers

Philippe Martinet, Univ. Blaise Pascal, Clermont-Ferrand (France)
François Chaumette, IRISA / INRIA, Rennes (France)
Koichi Hashimoto, Tohoku University, Sendai (Japan)
Ezio Malis, INRIA, Sophia Antipolis (France)

Aim and scope

During the last decade, visual servoing has become a well-established robot control technique, integrating vision in feedback control loops. The conjunction of fast and affordable vision processing systems, together with progressive theoretical developments has led to an increasing number of applications. The subject has become increasingly more present in conference tracks (ICRA, IROS, ICAR), and journal papers.

The aim of this tutorial is to clarify and present the most recent and advanced issues in visual servoing systems. Though the tutorial aims to present the principles of such techniques, attendees are assumed to have basic notions of image processing, robotics, and visual feedback control.

The choice of visual features has been a historical matter of discussion in the visual servoing community. New concepts about general yet powerful visual features will be introduced in this tutorial. The classical division between 2D and 3D visual servoing models has been mostly overridden by new hybrid approaches, furthermore coping with calibration issues.

Performance is a critical issue in visual servoing systems, where high-speed feedback and processing is required for real-time applications. New software and hardware techniques need to be taken into account for a successful implementation.

Last but not least, visual servoing is spanning a wide range of unforeseen applications, which are possible thanks to recent theoretical developments, and major improvements in equipment. Novel applications together with advances in classical domains will be presented.

Schedule

Lecture 1 : Introduction to Visual Servoing (Philippe Martinet, Professor in LASMEA, Blaise Pascal University, Clermont-Ferrand, France)
Duration : 40 minutes
Content :

Computer vision basis used in visual servoing (camera geometry, ...)
From dynamic to kinematic control of robots (the task function approach)
Visual servoing scheme (basis)
Overview on visual servoing (bibliography)

Lecture 2 : Modelling in Visual Servoing (François Chaumette, Director of Research at IRISA, Rennes, France)
Duration : 40 minutes
Content :

From image point coordinates to image moments
3D visual features
Modelling with omnidirectionnal features

Lecture 3 : Vision Based Control Laws (Ezio Malis, Research Scientist at INRIA, Sophia-Antipolis, France)
Duration : 40 minutes
Content :

Intrinsics-free visual servoing
Choice of Vision based control laws

Lecture 4 : Micro Visual Servoing (Koichi Hashimoto, Professor at Tohoku University, Sendai, Japan)
Duration : 30 minutes
Content :

Micro visual feedback (MVF) technology
High-speed vision sensors
Micro-size manipulation

Lecture 5 : Applications (All lecturers)
Duration : 30 minutes
Content :

Illustration of applications
Illustration of theorethical points
Visual Servoing Toolbox Initiative

Open questions and discussions : All lecturers and participants (free of time)

This page has been accessed times since 2004/06/25

Last modified by Enric Cervera
on 25/02/2010 @ 14:05.