Robo Home | TargetingChallenge2K7 | Changes | Preferences | AllPages

This challenge is in preliminary development. Anything about this challenge may change.

See the TargetingChallenge2K6 for general guidelines about participation.
See the TargetingChallenge2K7 for discussion of this challenge.

While reference bots are being determined (try-out phase) this listing will be in the style of a movement challenge.
Formatting may need some work as we get more challengers.

-- Martin

Bot & LinkAuthorType[Ugluk][Dookious 1.25][GrubbmGrb][Seraphim dev43][Che 1.2MC][Chalk 2.5.N.MC](challenger n)
[Grid 11.27]MartinRM36.89 4.74 10.24 4.5 14.70 5.68
[DuelistMicro 2.0 movement (non-surfer)]David AlvesRM12.36 7.9012.13 10.04 20.23 11.90
[FortuneHT 1.0]KevRM22.04 12.92 20.93 16.84 28.79 20.14
[FortuneRM 1.0]KevRM17.32 14.29 17.29 16.28 26.82 16.31
[FortuneGT 1.0]KevRM20.49 11.78 19.15 18.54 25.08 14.65
[Fortune 1.0]KevMM19.78 18.43 23.01 20.19 26.41 18.92
[GrubbmGrb 1.2.4]GrubbmGaitOSC19.17 8.06 24.79 14.14 28.97 11.66
[Cigaret 1.31TC] Iiley RM 15.23 10.89 27.63 10.79 32.70 11.28
[RaikoMicro 1.44TC] Jamougha RM 21.49 15.67 19.60 20.26 24.92 17.25
[Archer 0.6.6TC] Ph RM 17.40 11.68 15.16 17.08 20.68 14.68
[WeekendObsession 1.6.3TC] Simonton RM 19.10 12.30 14.56 12.94 24.60 12.71
[WeeklongObsession 1.5TC] Simonton RM 21.56 8.43 20.73 21.13 29.00 13.50
[Komarious_1.705TC]VoidiousWS30.10 12.38 47.35 18.2 35.13 13.25
[Dookious 1.462TC]VoidiousWS 39.81 24.92 44.87 37.98 38.39 32.77
[WeeksOnEnd 1.10TC]SimontonWS35.84 17.96 45.88 35.78 34.89 27.05
[Seraphim 0.052MC]Chase-san WS 30.76 22.22 45.86 28.46 30.60 27.55
[WaveSerpent 1.221TC] Kev WS 32.59 12.10 52.64 30.71 29.84 27.22
[Chalk 2.5.N.TC] Corbos WS/DC 38.08 36.60 49.28 35.37 37.37 32.17
[Shadow 3.66dTC] ABC WS/DC 39.36 32.32 40.79 34.13 33.41 36.42
[RaikoMX 0.32TC] Jamougha WS 29.67 15.39 49.66 25.52 32.80 28.24
[Cyanide 1.80.bTC] Alcatraz WS 25.49 21.86 51.06 23.30 30.46 23.97
[CassiusClay 2pi.08TC] PEZ WS 38.77 28.62 57.89 25.43 38.59 29.09

I updated DuelistMicro to make it a bit harder to hit. It's roughly equal to Cigaret now, a little bit easier. --David Alves

I get the same result, Ugluk page is missing. -- Chase-san

I thought I'd uploaded it, but it appears missing now. I'll make it available in 2 hours when I get in to work. -- Martin

It is interesting (to me) that the bots I have an easier time hitting are more difficult (relatively speaking) for Dookious. -- Martin

Added GrubbmGrb as challenger and as reference. Although its movement does have a random component, the main part is a double oscillator. -- GrubbmGait

If you wanna test vs Seraphim, you need set it to reference mode in its properties file. (This version actually got better targeting scores overall then Beta, e.g. 0.023) --Chase-san

I know it's a minor thing, but I think it would be useful if you could post a version with it already set to reference mode. It's so easy to do and saves anyone messing it up when they run it. -- Voidious

I didn't because thats the same one thats currently in the rumble, but I can make a seperate version for you. =) --Chase-san

Amazing GrubbmGrb score for WaveSerpent... -- Voidious

Yah, GrubbmGrb uses all the basic methods of firing in one big array, which is probably why its rating is so darn high in the rumble. --Chase-san

I think its expert movement against simple targeters is a very big reason, too, but his VirtualGuns are certainly a lot harder to surf than any individual simple targeter. -- Voidious

Thats what I was getting at, as it does well dodging all the simple methids, which GrubbmGrb uses, I need to refine my sentences more. --Chase-san

Simonton and Chase-san, would you mind posting links to your movement only versions that you posted the results for? Feel free to just mail them to me, if that's easier, and I'll post them. I may want to add another reference gun to this list sometime, so I'd like to be able to run it against all the current challengers. -- Voidious

Ahh, sure Seraphim just has the option changed in the properties file (as per normal) should be hard for me to change it. I'll get a copy posted soon --Chase-san

By the way, if Kev or Corbos would prefer, we can obviously withdraw WaveSerpent or Chalk, or change which versions are entered. I think those are the highest rated versions of each. That Dookious score for Chalk is just outstanding. -- Voidious

Hmm, GrubbmGrb would make a great movement challenge bot. --Chase-san

I am afraid you're reading the table wrong, the GrubbmGrb column just indicates that it can't hit a more advanced mover, as the surfers all score more than 40%. Voidious was right, it is ranked that high purely due to its performance against the lower half of the ranking. Just compare its details against any good or expert surfer and count the number of 90% scores it can equal or even beat the surfer. It is a good bot to test against in the beginning though, as it will exploit any flaws in your movement. As for reference in the TC, well, there are not much non-surfers volunteering to get slaughtered . . . -- GrubbmGait

Well, its hard to find good non-surfer movement. Genesis is gonna have a Minimum Risk Movement combined with Wavesurfing (I suppose its gonna be a little like TAOW, cept with surfing stuffies), that way it will be easier to integrate melee movement. --Chase-san

I have no problem with you guys entering WaveSerpent to the challenge, and I'm pretty sure that 1.221 has a better movement than my other versions (challenge-wise at least). If you guys are interested, I have yet another "Fortune" random mover. It uses go-to style movement kind of like Cigaret's, except that it's non-reactive. Of course, that would mean one author having 4 entries, so maybe I'll leave entering another non-surfer to someone else. -- Kev

Well, I think you should just post it in any case. I'm thinking we'll vote later to decide on the entrants - if you end up with 4 bots in there, so be it. :) And it could be better than one of your other ones, too, right? -- Voidious

We need a pattern matcher and maybe a DC gun as references. I suggest perhaps Simontons pattern-matcher as its doing so very well, however my computer is a little slow for pattern-matchers (i'll do it if noone else volunteers). --Chase-san

I think Che is the best PM to use for long battles, because it uses a FoldedPatternMatcher that doesn't really slow down. (That's why it's used in MovementChallenge2K6, too.) I'll gladly run Che and Chalk (and/or Shadow) against all the movers, but I do need a link to the WeeksOnEnd 1.10TC. Also, I think execution speed is very important when considering TC reference bots, so I may post some reference times on my machine for each movement. -- Voidious

Think you can edit Che so it doesn't lag at the end of the round after it runs out of energy? --Chase-san

Hmm, that doesn't sound too hard, it might just keep aiming continuously at that point. However, Che is only being used as a reference score for deciding this challenge's bots, right? So I'm not sure it's worth the time; plus I don't think it would affect total execution speed that much if you have it minimized. -- Voidious

Eh, whatever. --Chase-san

I really don't think Single Tick Pattern Matching is a good choice, either, for another reason. There are 2 bots in the rumble that use it (both mine), so it's hardly going to affect anyone's rumble scores to be good against my 2 guns. It actually wouldn't take much to trim down it's execution time by quite a bit (when codesize isn't an issue), and it also does not get slower with time. Actually, it should get slightly faster with time, after the pattern history has warmed up. Still not as fast as a folded pattern matcher, though, that's for sure. I'll post 1.10TC to the repository tonight. -- Simonton

If it keeps at this rate I might have Genesis' melee movement done in time for testing. I have a neat idea of using a line segment to allow the melee movement to not attempt to outrun the bullet, but instead move away from its path of flight (e.g. the line segment), weighted on the bullets power and nearness. --Chase-san

Isn't it pretty hard to track enemy bullets in melee? Or do you just mean in a 1v1 situation? -- Voidious

One on One situation of course, I attempted melee surfing, however the it lagged out, as I said before I am integrating melee and OneOnOne movement, and this is just another step towards that most totally cool merger, I was planning a really complex projections, angle limitations to keep the melee movement from colliding into bullets, but this is a great deal more simple (and workable). --Chase-san

I added results for WeekendObsession. I also ran WeeklongObsession, but the results are waiting for me when I get home, so I can't post them now. I don't know if anyone thinks this might be valuable, but maybe it's worth having a nano in the challenge? I allows you to see how well you can hit a bot that has no wall avoidance, no distancing, etc. I'll upload the jars to the repository tonight. -- Simonton

The 'problem' is then that it is more difficult to get a good score, as energy is lost when hitting wall and challenger. On the other hand, being quick to kill that referencebot would get you a better score, and being quick does pay of in the rumble. (Less time available to be hit yourself). I think the Fortune bots of Kev are already small (micro?) but the more choice, the better. -- GrubbmGait

I don't know, though, it might be a really good measurement of a gun to see how fast you can kill them. I remember Alcatraz once asked why it was easy for any bot to get 95-96% against Fhqwhgads but the best guns could get 98%, and David Alves said he thought the wall hits was the reason (on TargetingChallenge/Results). Good scores for WeekendObsession, though, seems like it's either hitting the walls a *lot* or not much at all :) I'll have RaikoMX and Cyanide results up shortly. -- Voidious

Are you sure you're not confusing Fhqwhgads with FunkyChicken? I can't say I know for sure that Fhqwhgads never hits walls, but I don't expect it would often enough to make a 2% difference in scores from it. -- Kawigi

Well, I actually don't know either of them all that well, I'm just basing it off of the two posts from Alcatraz and David Alves. He could be totally wrong about that being the reason, in which case I'm sorry for referencing it =) -- Voidious

Robo Home | TargetingChallenge2K7 | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited March 14, 2007 5:08 EST by Voidious (diff)