[robocup-humanoid] Rule Changes: Field of View

Thomas Röfer Thomas.Roefer at dfki.de
Mon Sep 17 05:42:58 EDT 2007


Hi,

> 1. If you have a robot standing facing the goal, it seems fair that the
> robot is able to see 90 degrees to its left and right, as well as 90
> degrees up and down.  This example does not violate the 180 degree rule
> it seems.
> 2. Take the same robot from example 1, and point the camera (or camera
> array) straight at the ground (imagine that the robot's body is
> transparent for this example).  You still have a hemisphere, but can
> see
> the entire field.  The field of view stops only at the horizon, which
> corresponds to the height of the robot.

This is a serious issue. There is the other one of an omnivision system with
a black 180° shutter rotating around the mirror at 25 Hz, so the robot can
take 25 images to its front and 25 to its back.

Actually, I find the +/-90° opening angle in any direction a little bit
much, because humans cannot see their feet when looking straight ahead, and
they don't see sharp in most of the area they perceive.

Best regards

Thomas Röfer

_______________________________________________________________________
Dr. Thomas Röfer                   Office Address:
DFKI-Lab Bremen                    Universität Bremen
Safe and Secure Cognitive Systems  Cartesium 02.054
Robert-Hooke-Str. 5                Enrique-Schmidt-Str. 5
28359 Bremen, Germany              28359 Bremen, Germany
http://www.dfki.de                 www.informatik.uni-bremen.de/~roefer

Phone: +49 (421) 218-64200
Fax:   +49 (421) 218-9864200
eMail: Thomas.Roefer at dfki.de
_______________________________________________________________
Deutsches Forschungszentrum für Künstliche Intelligenz GmbH
Firmensitz: Trippstadter Straße 122, D-67663 Kaiserslautern

Geschäftsführung:
Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff

Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes

Amtsgericht Kaiserslautern, HRB 2313






More information about the robocup-humanoid mailing list