Robo Home | Changes | Preferences | AllPages

WaveSurfer Gun Challenge: Compare your gun to the big guns in the Rumble!

With a common movement we should be able to determine which gun performs best in RoboRumble conditions. This time there are no restrictions on firepower. Unlike the RRGunChallenge, this challenge uses a simple WaveSurfing movement; it should provide a more accurate benchmark for how a gun will perform on a WaveSurfer.


Current scores:
Name Gun author Rating Date Remarks
Dookious 1.161 Voidious 2071 2006/09/14
Lukious 1.10 Voidious 2070 2006/09/14

Old scores, using Komarious 1.37 movement.
Name Gun author Rating Date Remarks
Dookious 1.10 Voidious 2001 2006/07/22
CassiusClay 2omikron PEZ 2001 2006/07/23
Ascendant 1.2.6 Mue 1992 2006/07/23
Phoenix 0.512 David Alves 1986 2006/07/28
Komarious 1.37* Voidious 1956 2006/07/22 Not a valid entry (it's a MiniBot), but included for reference.


By the way, I sort of consider this challeng to be in "beta" stage. If the rest of you guys find something wrong with it, or have a better way to implement it, we can certainly change it. -- Voidious

Wow, that's interesting... I hope I didn't mess something up in the implementation of Ascendant's gun. -- Voidious

"The resulting bot must be a MegaBot."... is that a mega version of Komarious? --David Alves

Nope, I just have trouble comprehending basic rules. :P My bad. -- Voidious

PEZ suggested, and I agree with him, that using CassiusClay's movement (or another top surfer) might be more helpful for this challenge. If we're going to benchmark how a gun works when paired with a WaveSurfing movement, we might as well use one of the top bots, right? Thoughts? -- Voidious

So, what do you guys think of changing to Butterfly movement for this challenge? -- Voidious]

Why? The purpose is to give an indication of how good a gun is when paired with a WaveSurfing movement. For benchmarking, I think any bug-free WaveSurfing movement is fine. Sure, using CassiusClay would result in stronger test bots, but I don't think it would result in more accurate indications of gun strength. --David Alves

I'm also not sure what the benefit of switching would be. It seems that if this is a consistent test of gun strength then changing to Butterfly will only shift the average score of the test bots up, not create a more accurate test. But, then again, I haven't even participated in this test yet so what do I know :) --wcsv

Consider what sets this challenge apart from the RRGC. Then it's quite natural to go for real WaveSurfing instead of some approximation that happens to fit in a mini. Not that I have looked inside Koomarious? but I know a bit about what it takes to fit both a good gun and WaveSurfing in a mini. Really good guns picks Koomarious? apart, no matter how strong gun you fit it with. From what I can deduce the strongest movement yet is that of Dookious. We could consider that too. But CC's movement has proven to be quite easy to plug into other bots. -- PEZ

That's an excellent reason why using Butterfky would make our test bots a lot stronger. But that doesn't mean that we're getting more accurate benchmark results. Especially since some of the more complex aspects of the movement (like Butterfly enabling or not enabling its flattener against a given opponent) could cause the score to vary quite a bit. I would expect less variance in the results using Komarious's movement since it is so much simpler. -- David Alves

Yeah, it makes sense. I still think CC movement would be a choice more in line with the goals of this challenge, but I'm not too sure I'm rihght so I'll rest my case. But, we should at ensure that the KomariMover class is MegaBot size to start with. -- PEZ

(Edit Conflict) Well, one thing to consider is this: I think we'd all agree that a "good", bug-free WaveSurfing movement and a decent gun should easily crack the 2K barrier; so far, only the very strongest guns in the rumble can crack 2K with the Komarious movement. I think her movement is "bug-free", but it is pretty barebones and far from the most polished WaveSurfing movement out there. It is true that Butterfly is more complex than a basic WaveSurfing, but I'm not sure if that would result in inconsistent score. There is also the option of removing the flattener from Butterfly for this challenge, or using RaikoMX or Cyanide instead. -- Voidious

For the record, RaikoMX uses a flattener too. -- PEZ

Well, how about Butterfly with the flattener (and any other MultiMode stuff you might have in the movement) disabled? I think the only real problem with it is the MultiMode aspect. I know for sure that Phoenix's results against other bots can vary wildly depending on if / when it decides to flatten or decides to use the anti-surf gun. --David Alves

That seems good to me. I think KomariMover is a lot better than RaikoNMT? for benchmarking a gun for a WaveSurfer, but I do think Butterfly would be better - at least for me. If/when everyone agrees with that, I'll gladly do the dirty work on Butterfly if PEZ doesn't have the time. -- Voidious

Yeah there probably aren't too many false positives for that =P -- David Alves

Ok, I commented out the flattener and random visits from Butterfly, called it ButterflyWSGC?, and posted the test entry for Lukious last night. The score shocked me a little, but that's OK =) For now, you can grab LukiSaberWSGC? or DookiSaberWSGC? and extract it for the code, but I'll post everything later today to make it more readily available. Oh, by the way, if nobody noticed, I did find a bug worth a few points in Komarious' movement recently. -- Voidious

I'd have participated but to include Komarious you are agreeing to the open source code condition of your resulting robot. -- Martin

Hmm... Technically, I suppose you are right about that. But I don't think anyone would mind if you posted an RRGC or WSGC entrant for Ugluk. I certainly don't mind. I know that ABC and Mue have posted RRGC entrants, and they don't open-source their stuff. Note that we are switching to this Butterfly movement, though, which you can find in my WSGC entrants or wait for me to post the files for later. -- Voidious

I haven't tried linking to a jar file other than Robocode.jar, but perhaps for challenges like these the piggy-backed code could be distributed in this fashion (as a jar to link to). -- Martin

Pretty sure we could do that, I'll look into it. -- Voidious

Where can I find the needed stuff (ButterflyWSGC or DookiSaberWSGC), so this challenge could be resuscitated. (and I need some 'real life' proof what I should do next) -- GrubbmGait

Doh, sorry I dropped the ball on this. You could get the files from extracting [DookiSaberWSGC 1.61.bfly], but I'll try and clean it up and write some reasonable instructions sometime soon. -- Voidious

Well, I managed to get re-acquainted with ButterflyWSGC? and got it setup cleanly for DookiSaberWSGC? 1.161. There was actually an issue with the above DookiSaberWSGC? - CC's movement depends on the bot extending the RumbleBot? class, using a RumbleBot?.enemyIsRammer?() static method for anti-rammer code. My bot wasn't extending RumbleBot? before, but now it is... I think many people do something similar, so having a WSGC entry need to extend RumbleBot? seems like a bad thing, right? I'll need to refactor a little more to change that. Hopefully I can finish it up tomorrow. -- Voidious

Argh, I forgot about this again! I'll try and remember to take another look tomorrow. -- Voidious

Robo Home | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited February 2, 2007 7:08 EST by Voidious (diff)