Hi!
Starting from RoboCup 2006 in Germany, there will be a new competition
held within the RoboCupRescue simulation league, namely the Virtual
Robots competition. This competition aims on simulating teams of real
robots within buildings based on the USARSim package that extends the
Unreal Tournament 2004 software.
Everyone is invited to participate in this league and to hopefully win
one of the prizes!
Cheers,
Alex
attached mail follows:
Dear All,
please find the tentative rules for the 2006 competition. Please
note the deadlines. A pdf version of the file can be found on
http://www.faculty.iu-bremen.de/carpin/VirtualRobots/
If there are comments, requests and so on, please post them to
the mailing list.
Best regards,
Stefano Carpin
------------------------------------------------------------------------------
The following is a set of tentative rules for the Virtual robots
competition.
Whenever possible, rules similar to those found in other
leagues/competitions
have been used.
1- Participants agree to release their code to the scientific community
after the competition. The winning team and authors of inspiring
implementations/algorithms will be required to make their code
available to the other teams on Sourceforge.
2 - In order to offer each team the same opportunities in terms
of virtual robotic platform, models that are intended to be used during
the 2006 competition have to be published by December 31 2005.
Publishing means checking the code in to the CVS repository on Sourceforge.
Basic documentation must be provided and added to the documentation area on
Sourceforge.
After December 31st, only bug fixes are allowed. This in particular
means that no additional commands will be added after the deadline.
Robot models must be consistent with state of the art robots. The
submitted robot
models will be evaluated for inclusion into the USARSim competition with
a final
decision being made by January 31, 2006. Compiled versions will be
released with
future USARSim releases.
3- For each robotic platform a maximum weight capacity will
be specified, while for each sensor a weight will be provided.
Developers will be
allowed to place any combination of sensors onto their platform. However,
exceeding the weight capacity of the platform will have a detrimental impact
on the team's final score.
4 –Teams are encouraged to preregister by Jan 15 2006. Preregistrations
should be sent via email to s.carpin AT iu-bremen.de (see
instructions on www.faculty.iu-bremen.de/carpin/VirtualRobots). According
to the size of this response, teams competing at RoboCup may be limited to
those already preregistered or satisfying additional qualification
procedures to join the roster
5- During the competition each team will be required
to run multiple times. The precise schedule depends
on the number of teams and will be anounced duringthe
competition. Different runs may be run on different
scenarios (arena). As a matter of principle, dynamic
environments may be used.
6 - During the competition the UT2004 based version of
USARSim will be used. The official release can be found on
sourceforge.net/projects/usarsim
7 - Before the competition each team has to provide their
ini file to be used during their runs (to speed up the setup
stage)
8- The organizers will provide the UT server and the
visualization client only. Each team is required to
bring their own laptop(s) to compete. Their laptops will power
the robot algorithms as well as their user interface. Wired ethernet
will be used
9 - Before each run, every team has to specify how
many robots will be used, and will receive in turn
a text file containing the start location for every
robot (x,y,z, an orientation)
10 - During the competition, teams will be allowed to
exchange as much information as they want between robots (this means via
TCP).
11 - Messages containing ground truth data will not
be sent by the server at any time during the competition
(this affects in particular the STA type messages). Noisy INU data will
be provided.
12 - As far as possible, an automatic refereeing system
will be used. In case this is not completely ready
and stable by the competition time, certain aspects will
be evaluated by an external human referee. This later
aspect will probably be limited to only map quality evaluation.
The competition will be won by the team scoring the
maximal amount of points. A scoring function similar to
the one used in the Real robot league will be used. The
following items will increase the obtained score
+ victims identified (location, signs)
+ map quality
+ percentage of the environment explored
The following items will instead negatively affect the
score
- number of operators
- bumps into walls/objects and into victims
- sensor overloading (see point 3), intended as exceeding
of the robot payload by excessive sensorial capabilities
Tentatively, a formula like this will be used
VICTIM_SCORE - K_1*BUMPS
[ ---------------------------------------------- + EX* MAPQUALITY ] *
K_ARENA
(1+N)^2
- The constants K_1 and K_ARENA will be fixed, and will be dependent
on the arenas used.
- EX is the term scoring the amount of the environment explored and will be
automatically determined by logging the path followed by each robot (and
then
compared with the overall environment extents).
- MAPQUALITY is composed of two terms which are explained in point 12
- each victim will carry an RFID tag that can be read at close
distances. Reporting
the tag information will be considered enough to identify the victim.
However teams
that can report additional information by using image processing will
receive
additional points
13- In order to assess the quality of the map, metric quality and
topological quality
will be evaluated. For metric quality,
RFID tags will be randomly scattered in the environment. Each tag
will be identifiable only once by each robot (this to prevent their use for
localization). Teams will be requested to provide a list of the
identified tags
an their respective placements (x,y,z). This information will then be
matched
against the known ground truth and the (mis)match will determine the metric
map quality score. For the time being, topological quality will be
determined
by the human referee
14- At the end of each run, each team will be requested to deliver the
following
information
a) victim information (RFID tag and victim position at least, any other
information
on top)
b) a text file containing all the identified RFID tags and their
positions. In case
multiple robots are used, a single file containing all the identified
tags will
be enough. The precise formatting data will be released at the same time
as the
automatic referee
c) a map of the environment in electronic format (preferably jpg or
ppm), for
topological evaluation.
All other information will be automatically gathered by the refereeing
system.
15- Information on the RFID tag sensors (generously provided by Alex Kleiner
from Freiburg University), will be introduced in the next USARSim release.
-- Dr. Stefano Carpin Assistant Professor of Computer Science School of Engineering and Science International University Bremen -- Germany _______________________________________________ robocup-rescue-v mailing list robocup-rescue-v@mailman.cc.gatech.edu https://mailman.cc.gatech.edu/mailman/listinfo/robocup-rescue-v
_______________________________________________
robocup-rescue-s mailing list
robocup-rescue-s@mailman.cc.gatech.edu
https://mailman.cc.gatech.edu/mailman/listinfo/robocup-rescue-s
Received on Mon Nov 14 15:49:43 2005
This archive was generated by hypermail 2.1.8 : Mon 14 Nov 2005 - 14:49:44 GMT