Re: Extended deadline, new version of Webots & new contest

"<[email protected]>" <[email protected]> Wed, 31 Mar 2004 13:46:09 +0200 (CEST)
Newsgroups gmane.comp.cybernetics.webot.contest
Message-ID <16251028.1080733569535.JavaMail.www@wwinf4005>
Hi everybody !

First of all I’d like to congratulate and thank Olivier for this great contest. It is very open and getting more and more exciting, I really enjoy it. 

I’d like to share my views about different issues Olivier talked about in his last mail.

“ I would like to release a new version of Webots to be used 
for the contest which fixes a number of issues with servos (…) So, this might have a slight 
incidence on the behavior of some robots”

There is only one month to go before the end of the contest, and I think it is almost a bit late to change the physics of the game again. Preventing impossible moves is a very good thing, but honestly I would prefer to spend the last month of the contest on working about strategy and image processing, rather than on rebuilding my robot movements a third time. Of course I am ok to adapt my robot to the new conditions if necessary, but I think that at some point ( preferably very soon )  we should all agree about a physic for the game and also about the rules, and not change them until the end because there is not much time left. After that any change/experimentation will be very welcome at the beginning of next year’s contest ;-)
 
 
” (2) The robots won't be allowed to fall / recover more than three times.”

I do not really understand the purpose of this rule: I think it is much more interesting to try to push the opponent out of the ring or to try to prevent him from recovering by kicking him where as he is still on the ground.
Will a dive be considered as falling on the floor? If no, how will the supercontroller make the difference? If yes, then the robots will not have many different strategies left and in my opinion the game will be less interesting:  the robots will only walk, turn, and try to kick their opponent with their arms, as any other kind of move or attack would be suicidal.
It also happens very often that the robots fall at the same time when there is a collision. What will happen if the 2 robots have already fallen twice on the ground and they fall again at the same time? The first one to be under the threshold has lost? This will only be a question of milliseconds…. and it is not necessarily the one who did the attack that reaches the floor after the other. And also it will not be very important anymore to be able to localize your position on the tatami, which looses some interest.
Finally, I think this rule would result in very careful robots that would no take a lot of risk, and also a lot of very nice moves like Capoeriste’s or Astalavista’s gymnastic “wheel” (is that the right english name?) earlier in the contest would not be implemented anymore.

” (4) The 3D view will display more information (like the 5 seconds count down 
when a robot is down and the result of a match: KO, time over, best 
distance, etc.)”

I think this is really cool, I specially really enjoy the idea of having the 5s count down displayed on the screen, because it is often not obvious to see if a robot is under the threshold or not. It would be really nice I think if it could be implemented before the end of the contest ;-).

”At this point many issues are open, I really appreciate your input to 
improve the organization of the contest. “

I’d like to propose another idea for a change in the contest:
Last year in the last week of the contest one robot was at rank 1 the 3 days before the end, and could beat nr 2 most of the time. But on the last day nr 2 won over nr 1 by 3 sets to 2 (without any update in the controllers) This is just to say that I think that there is still a part of random when 2 robots compete in 3 winning sets, and the hall of fame is not very stable even when the competitors do not update their controllers.  
Of course this is part of the game, but I think it would be much better if the robots competed in many more sets, like for example 10 winning sets. This would give a better idea of the relative strength of the robots. I agree that  it would be very long and boring to see a match Goldorak against SmartGhost in 10 winning sets, but I think that it would be a very good idea to increase the number of sets for the top ten robots only for example. 
 The day’s round would be a bit longer but who cares? And in the top ten there are no static robots, the games are very interesting and they usually last very less than 2 minutes, so it would not be so long, maybe one more hour(?). The .wma record files would be bigger and a bit slower to download, but this would also be the opportunity to have more feedback about the behavior of the robot on the contest server (which is a bit different than on a local machine, at least for me). What do Olivier and other competitors think about that?

Well, this was just my point of view, and of course I think the best rules are those that satisfy the most people. What do you think ?

Good luck to everybody !!

Xavier ( Kubiak )


------------------------------------------

Faites un voeu et puis Voila ! www.voila.fr 




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
     http://groups.yahoo.com/group/webots-contest/

<*> To unsubscribe from this group, send an email to:
     [email protected]

<*> Your use of Yahoo! Groups is subject to:
     http://docs.yahoo.com/info/terms/