[robotics-worldwide] Tutorial on Real-Time Computer Vision for Localisation and Mapping

W Mayol-Cuevas wmayol at compsci.bristol.ac.uk
Tue Nov 3 15:23:08 PST 2009

As part of IEEE Humanoids, on the 7th of December 2009 in Paris.
Early registration for this tutorial is about 30Euros or less before Nov

Andrew Davison (Imperial College)
Andrew Calway (Univ of Bristol)
Olivier Stasse (AIST/ISRI-CNRS/STIC)
Walterio Mayol-Cuevas (Univ of Bristol)

For a highly dexterous and agile platform such as a humanoid robot, a
pre-requisite for efficient task or motion planning is to be able to know
where it is relative to its surroundings. In recent years, there has been
significant effort in developing real-time mapping and localisation using
vision sensors only. The attractiveness of visual sensors is not only
their size, low cost and availability but importantly the different kind
of features they can recover.

Moreover, a humanoid requires fast algorithms able to provide pose within
a few milliseconds and therefore the visual techniques required must be
robust and consider real-time constraints.

This tutorial will concentrate on the following topics:

-Basics of Visual Localisation
-Introduction to Visual Mapping
-Real-time and robust feature matching and description
-Creating larger maps and discovering map structure
-Navigation and visual sensor planning


Registration to IEEE Humanoids, this tutorial and more details are
available at:


More information about the robotics-worldwide mailing list