Archived talk:User:Tango 20090519
- Tango Sub-pages:
- Tango - Archived Talk 20090519
|
News Flash: I'll be uploading a new bot tomorrow, it's not brilliant, but its much better than all the others i've made. I've been testing it today, i just need to come up with a name and i'll upload it. It was just a test bot to see how bullet dodging worked (hence its working name: Dodger), but it turned out to be better than i expected once i added a simple gun.
Cool! I suggest you name it Cassius after Cassius Clay (later Muhammed Ali). He was an excellent dodger with a remarkably effective punch. He can also provide your bot with a punch line (unavoidable pun): "Dance like a butterfly! Sting like a bee!" What kind of simple gun are you using? -- PEZ
In fact, now when I have started to think about it, Robocode OneOnOne battles and boxing have very much in common. It's all contained in a rectangular arena. It's about the perfect combination of dodging and hitting. It's about power management. It's either about knocking your opponent out or making points. We could go on. You have a deep name space to pick names from as well. And all these boxers have produced more than one tag line you can use as well. -- PEZ
Yeah, like Mike Tyson- "I can eat your ear, because I'm supposed to be the champion." -- Kawigi
Nice idea, but i'm not really a boxing fan, so i'll come up with something else. Thanks! -- Tango
Ok, i've thought of a name, its not very good, but then the robots not very good so it fits nicely. ;-) The robot is now called: Recrimpo! For more details see that page. -- Tango
Hi! I'm a complete novice to Java and Robocode, i have made a very simple robot, TangosBotV1, which will beat most of the sample robots but little else. I'm planning to make a robot using VirtualBullets and AntiGravityMovement as soon as I have the knowledge to do it, this robot is Deltan.
Considering very view coders are working on dedicated team robots, i have decided that i will. I think this will give me a chance to do well, because there are less unbeatable teams out there. (A team of 5 times SandboxDT is a daunting challenge though...) I don't have a name for my team yet, any ideas will be welcomed. It will be designed so i can change how many of each type of robot there are, and it will still work. It will be focused around a shared hash of all enemies, all the robots with radars will add to it. This gives a lot of scope for pattern matching, if only i knew how to do it...
Welcome to Robocode land and to this wiki! I hope this wiki will help you come up to speed quicly with your next generation bot. My spontaneous advice when I see your plans forms a new section on the Beginners page. -- PEZ
Thanks! I've just started work on my new bot, Deltan, so I'll might have a few questions to ask soon. What is the best place here for random questions? -- Tango
Anywhere. Most often you can use the search function to find a place where the question fits best. And if not use the front page or the Requests page. -- PEZ
Were you planning on putting Recrimpo or any of your Haikus you posted into the RobocodeLittleLeague? -- Kawigi
Maybe... they will lose... umm... ok, here's what i'll do. I'll ignore recrimpo because it is a terrible bot, and i will post a competative haikubot (the ones i've posted so far have been attempts at copying other bots, rather than real attempts to win battles) either tonight or tommorrow. How does that sound? -- Tango
Sounds great. I figure tomorrow night will be the deadline for season 1. -- Kawigi
Ok, Tango i have an idea. Having seen the way PEZ and Jamougha have urged each other on, I've decided that i need a little head-to-head competition. I propose some kind of 2nd place brit competition, something along the lines of the old PEZCrippaLeague. Every week we have a OneOnOne battle to see who's the winner. What do you say?? -- Brainfade
I'm on half term at the moment, so I should have time to fix whatever is wrong with my GF gun. If I can do that, then I might be able to challenge you. What place is your top bot in the rumble so far? Are you thinking megabots, or size restricted? My current bot is mega, but I am considering getting it down to mini if I can get it to work. -- Tango
Interesting. The PEZCrippaLeague really kept Crippa and me on our toes for a long while of our early Robocoding. Yes, find someone competetive around your own current level and start competing. That will surely accelerate your advancement. Another thing to remember is that Robocode competition below the top 2 bots are still rather simple. Raiko proves this. Bug frequency is a major parameter I think. And to get that I can very much recommen tests. Got a GuessFactorTargeting gun that performs sub 85% in the TargetingChallenge? It probably has a bug or you're not segmenting it very well. -- PEZ
I don't have an entry at the moment, and i'm keen to get one sorted asap. I would much prefer Megabot's as i've never tried restricted codesize, and my code is really wooly. Besides, the bot i'm working on is way over a mini and it doesn't work yet. If you've got some time, work on your bot, and give me an answer when you've decided... --Brainfade
Well, I've found the problem with my bot, I just need to work out how to fix it (I've posted the problem on the GuessFactorTargetting/Tutorial if you want to offer any advice). It's a very simple bug, so I should have a working bot very soon, so I will take you up on your challenge. May the best bot win! -- Tango
Well, now you've got a working gun, i'd best put some graft in. I'm gonna need this week to put it together (I'm a notoriously slow worker) so shall we schedule teh first run for next week?? -- Brainfade
How about next Sunday, then I have the whole half-term to work on it (or should I be doing homework... nah)? I'm just running my new gun through the TargetingChallenge, so we'll see how well it works... -- Tango
79%.... according to PEZ that means there is something wrong... it could just be the segmentation (simple distance segmentation at the moment), or it could be annother bug... I'll put it together with some movement, and see how it fares in the RR@H. I guess it can't do much worse than the current version... -- Tango
That got to 148, but i've added data saving now, so hopefully it will go up more. -- Tango
79% with distance segmentation only sounds like it is working. -- PEZ
That's what I thought. Once 2.51 has stablised, I'm going to start testing some segmentation. My first thing is going to be an aceleration/deceleration axis, i think. I'm not quite sure why, but i've got to start somewhere, and my gut feeling is that will be quite useful. -- Tango
Hey Tango, just curious, you've been here for a long time and you're very active, so why do you only have one bot?
Because I'm one of the worst coders known to man! I am working on a bot that seems to be working quite well. I might even release it this week. -- Tango
- [View source↑]
- [History↑]
