[robocup-legged] Rule Changes and Poll

[Walter Nistico]

> 1)
> Who will argue for continuing the 4L if the last year of the competition
> with Aibos was neither high quality nor entertaining?
-------------
Actually the discussion is between innovation and game quality (or
entertainment).
If we approve a series of important changes to the rules, innovation is
granted, but game quality and entertainment are not.
However, honestly nobody can say that the humanoid league, as it is now,
it's entertaining, yet it is the most interesting, challenging and
innovative.


> 2)
> For a major innovation leap forward there are more appealing
> and more challenging alternatives
> than discussing beacons/localization only, for example, a very new
> tournament structure allowing only soccer games of joint teams of the
> software of 2 teams in 2007 (take, e.g., the NUbots and UNSW demo of 2005
> or the 11-11 demo game in 2006).
> The round robin could be played in a variation of combinations
> e.g., teams A & B vs. teams C&D, then A&C vs. B&D and so on
> and each team getting their joint score as individual record
> for determing the round robin ranking.
>
> Therefore the field and team size could be increased as well,
> e.g., field in length and breadth times 30% and teams of 6 players.
>
> Such a change would be a real major innovation which other leagues
> never thought of implementing yet which would show the 4L on the cutting
> edge of multi-robot research.
>
> The 2007 year of change would be an ideal date for some more revolutionary
> changes. I am sure there will be many more exciting ideas
> for major innovations in rule changes which would make the games in 2007
> more than a localization challenge with no beacons.
>
> I believe the 4L should try a more revolutionary step which would also
> make all other leagues very curious to watch the games in 2007
> (the change of beacons will not).
----------------------------------
Actually the TC is discussing several other possible rule changes.
However, the proposal of "mixed teams" poses a series of important
problems: first it should be decided whether the mixed teams are
statically or dynamically linked, so to speak.
With statically linked, i mean something like the GermanTeam or TeamChaos,
where a certain number of teams join forces and jointly develop a common
code.
With dynamically linked instead, i mean something like "pick-up soccer",
where the robot code is developed independently from the various teams,
but it must use a certain communication and coordination protocol so that
anybody can play together with anybody else.
Concerning the static team linking: i do not think that forcing a
cooperation among teams is the right way to go. Such a cooperation poses
already several important challenges when it is voluntary, and a
rule-imposed cooperation could easily turn into a total disaster.
Concerning the dynamic team linking: first off, there are too many
possible permutations to generate the mixed teams as suggested, and this
would result in an impractically high number of games.
Second, the game quality would suffer greatly and really retrocede the
league to pre-2002: in each match it would be enough that only one part of
one team does not cooperate correctly to screw up the whole team play
completely (e.g. one robot always runs to the ball); also it would be
really difficult to devise real team strategies beyond a simple "area
based" (as used in the 11 VS 11 match).
Even in human pick-up games, the matches tend to greatly emphasize the
1-on-1 skills rather than the team play skills.
At last, the results would be too much affected by random factors, like
how well a certain combination of teams works compared to another one etc.
well beyond the relative skills of the sub-teams involved.
Greetings,
Walter

--
Ing. Walter Nistico
Robotics Research Institute (IRF)
Information Technology Section
Dortmund University
Otto-Hahn str. 8
44221 Dortmund, Germany
Phone: +49 231 7554533
Fax: +49 231 7553251