Talk:Challenge 2K9

From Robowiki
Jump to navigation Jump to search

Note for the reference gun/movement:

  • The RaikoGun is almost as same as MC2K7, but under difference package and have small tweak.
  • The KomariousFlatSurfing is Komarious-based wave surfing that surf like flattener always on and have better distance controller. (Dookious is suppose here)

» Nat | Talk » 13:18, 14 March 2009 (UTC)

TC Fire Power Discussion

I didn't find (maybe I didn't search enough) any discussion about the fire power in the targeting challenge, but I think that making the bots fire a fixed bullet power is buggy, I'm sure many guns perform better with some powers (I know YersiniaPestis gun do) and I'm not talking just about gain/energy or damage/energy, but my hit rate can drop over 50% using 2.5 bullet powers instead of 1.9. Those values I tested, but I'm sure there are many more differences with other powers, and I think similar effects can happen to many guns. And besides that, the energy management may be part of how good a gun is, I use low powered bullets when my hit rate is low or have too few factors recorded so it can learn faster. And probably there are guns that increase fire power according to their hit rate. --zyx 01:30, 24 March 2009 (UTC)

Fire power of ??? Reference or Challenger? For reference, I'll leave it alone (not to touch it.) like MC. But for challenger, I know that almost all robot fire at 1.9. There no discussion on this. I just decide to use 2. There are nearly no difference between 1.9 and 2. Actually, I want 2.5 but nearly every robots fire at 1.9 so... If I let user choose fire power, it is not challenge anymore. » Nat | Talk » 02:37, 24 March 2009 (UTC)

  • I meant challenger. Mostly I wanted to know if there was a discussion about it, I personally would prefer to let the challenger use it's own, but in general I'd like it to be discussed before it is ruled. If most people agree it should be a fixed value then majority rules :-). But I strongly believe that the fire power is part of the gun, I usually run several targeting challenge seasons only testing fire power because different values give me very very different hit rates. --zyx 02:50, 24 March 2009 (UTC)
  • Then let start discussion here. Every old TC use fixed power of 3. I know really that it reduce hit rate. For now, I vote for between 1.5 and 2.5. Lower than 1.5 is too low. » Nat | Talk » 03:24, 24 March 2009 (UTC)
  • Yes I know old TC were fixed at 3, but this a new year :). I could agree with a range as you say. Although I would prefer a lower cap only, 1.5 sounds good but I wouldn't cap it on the high side, my bots rarely fire 3 powered bullets, specially if the distance will be controlled by Komarius and not me, so 2.5 will not make much difference for me, but I know there are top bots that do use 3 more often. --zyx 04:43, 24 March 2009 (UTC)
  • Actually, TC, TC2K6 and TC2K7 all fixed firepower at 3. Unlike original Komarious, my modified version of Komarious will control the distance at 450. I can said that it is common distance. DurssGT control at 400, Phoenix control at 475 (+ dynamic distancing), Dookious control at 465. Shadow distance is quite close, but not less than 300. » Nat | Talk » 07:58, 24 March 2009 (UTC)
  • Well, while I agree that fire power is a meaningful part of the gun, I'd rather not bring it into place in the targeting challenge, because that ruins the usefulness of the challenge for gauging one's adaptive guns. Maybe it would be reasonable to make a "secondary" results table that allows any fire power scheme to be used, but I very strongly think that the uniform firepower between bots is very important. One thing that I would say wouldn't ruin that gauging however, would be if firepower was non-constant, but every single bot had to use the same algorithm to decide it. --Rednaxela 07:17, 24 March 2009 (UTC)
  • If we're only allowed to shoot at 1.9 it is not at all realistic against, for example, Rambots. If you want to try this tactic you will get demolished. I agree with Zyx that we should be able to vary our bullet power, as this is definitely an integral part of the gun, that decides it's effectiveness in the rumble. However, if we all take the same tried and tested bullet power algorithm from one of the top rumble bots, that would also be fine. Something like
    power = distance<300?3:Math.min(Math.min(myEnergy/16,(enemyEnergy+0.09)/4),1.9);
    --Skilgannon 14:36, 24 March 2009 (UTC)
  • I never thought about all having same good algorithm for bullet power, but now that is mentioned I believe is the best and most fair solution :). --zyx 14:54, 24 March 2009 (UTC)
  • I always use this solution:
    limit(0.1,distance < 150 ? 3 : Math.min(Math.min(getEnergy() / 3, enemyEnergy / 4), 1.9),3)
    in all my (mega)bots. I adopt it from DCBot, really. Here is what out king in March 25, 2009 done, simplified:
    power = Math.min(Math.min(enemyEnergy / 4, myEnergy), energy < 5 ? 1 : (energy < 15 && enemyEnergy > myEnergy) ? 1.4 : (energy < 25 && enemyEnergy > myEnergy) ? 1.59 : 1.9 );
    and this is what out current king do:
    power = getEnergy() > 1.9 ? 1.9 : 0;
    I think I'll use current king way of choosing bullet power. =) Anyone know who is first use 1.9? » Nat | Talk » 04:00, 27 March 2009 (UTC)

I now come up with simple solution. How about this one?

e.getDistance() > 100 ? 2 : 3

This should still hold the 'controlled environment' and still can fire hard at rambot. » Nat | Talk » 16:12, 23 April 2009 (UTC)

Postponed to October

Sorry guys. My school semester is now open and I have a lot of school works to do. I'll have to postponed this to October. If you want to test your robots before that time, actually almost all discussions here is concluded. I'll need to modified a large set of modified robots, which I think you can do it except for PM challenges and WSCBotDEGF. Sorry again. » Nat | Talk » 15:47, 27 May 2009 (UTC)

I hate to be a spoil-sport, but this "challenge" seems very bloated and unweildy. What's gained by putting all of these challenges under one umbrella instead of creating them one at a time? When you want to benchmark overall performance, that's what the rumble is for. =) --Voidious 15:29, 10 July 2009 (UTC)

I just follow the TC2K6 way. But I've consider for few months now that this is really messy. I should re-consider this challenge. But I think that the most important one is SmallBot challenge, we really don't have challenge for them. I may merge all other challenges together. » Nat | Talk » 15:32, 10 July 2009 (UTC)

And I agreed with you, it is really very much bloated and unwieldy. =) » Nat | Talk » 15:34, 10 July 2009 (UTC)

There are no threads on this page yet.