TITLE: A Fast Vision Sensor Model: Matching Edges with NightOwl

PRESENTER: Raymond Sheh, http://www.cse.unsw.edu.au/db/staff/info/rsheh.html, rsheh@cse.unsw.edu.au

AFFILIATION:PhD Student,

DATE: Friday 3rd December 2004

TIME: 10:00:00

PLACE: K17_113

ABSTRACT:

Accurate localisation is a key requirement for most autonomous
robots. Increasingly, mobile robots use camera sensors. We describe
an algorithm that uses edge features in a visual image to implement
a sensor model for a given environment. The algorithm (NightOwl) was
developed to help localise a Sony legged AIBO robot on a RoboCup
soccer field, but has more general applicability. The NightOwl
sensor model is optimised for efficient use of computing resources,
is effective even when edge features are partially occluded and can
be used with a variety of Bayes filter localisation methods.

BIOGRAPHY OF SPEAKER:

Host:

Seminar Convenor:

Van Hai Ho