Difference between revisions of "User talk:Robar"
(24 intermediate revisions by 5 users not shown) | |||
Line 1: | Line 1: | ||
Please write about Adaptive Oscillator Movement and your new micro bot, please? I'd like to know your ideas. Also, do you want to race with me? I'm 13 years old. Do you want to have some race to top nano or the race into The2000Club? I'm not nano specialist but you are, let's try! » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 02:01, 15 February 2009 (UTC) | Please write about Adaptive Oscillator Movement and your new micro bot, please? I'd like to know your ideas. Also, do you want to race with me? I'm 13 years old. Do you want to have some race to top nano or the race into The2000Club? I'm not nano specialist but you are, let's try! » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 02:01, 15 February 2009 (UTC) | ||
− | Ook, I'm on it. ;) AOM is strongly under construction, but one of my new nanos will use a more sophisticated version of it. Currently Pugio moves somehow like that. It's about changing the frequency of oscillation in order to find a good solution against hostile guns. However, I'm working on my new nano's movement | + | Ook, I'm on it. ;) AOM is strongly under construction, but one of my new nanos will use a more sophisticated version of it. Currently Pugio moves somehow like that. It's about changing the frequency of oscillation in order to find a good solution against hostile guns. However, I've put off AOM recently, because I'm working on my new nano's movement. |
− | You mentioned my new microbot. Actually, I have two, but they are not so good at the moment, because I'm too lazy to learn GF. They use almost the same gun that is in my nano's! The difference is that the micros match velocity | + | You mentioned my new microbot. Actually, I have two, but they are not so good at the moment, because I'm too lazy to learn GF. They use almost the same gun that is in my nano's! The difference is that the micros match velocity instead of lateral velocity. This makes their gun inaccurate, but it doesn't get mad by their random movement at least... --[[User:Robar|HUNRobar]] 17:00, 15 February 2009 (UTC) |
+ | |||
+ | GF seem to be hard, but it heart is very simple! At first, I'm too lazy to learn it. But once I've read more on robocode bot/ranking. The only way to get in top10 of general roborumble is to learn GF or learn TronsGun with DynamicClustering. I prefer first choise at that moment. (Now, I understand both) Once I've got the heart of GF, no problem. I take 2 days understand heart of GF and 2 more weeks to understand how to use rolling average with GF! Once a while, I've clean up the GuessFactor Targeting Tutorial from the old wiki, you can start on that! Feel free to ask me or someone on this wiki for your question. | ||
+ | |||
+ | Another thing, you should go to the old wiki often. There are much more rich information to acquire on the old wiki. One more: Could make your nanos open source, please. I know I can decompile it but almost every nanos release their source code! The robocoder here will give the credit, of course... Ah I think you use IDE, right? is your IDE split up bin and source folder? I think I saw your problem. Now, I've same source and bin directory in order to package robot with source. You may done this, too. Or can can post your code under suppage of your bot page. OR you want them close source? That bad for beginners! (I think I'm enter robocode community later than you, but, sorry, I think I know more ''theory'' than you.) | ||
+ | |||
+ | » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 17:37, 15 February 2009 (UTC) | ||
+ | |||
+ | Don't know, I always tick the include source box. It doesn't include source for some reason recently. | ||
+ | I looked up and yes, I've set .../build/classes/ to load my robots from. It means only class files are seen by Robocode. But I don't know how to make Netbeans generate the source into the build folder. --[[User:Robar|HUNRobar]] 17:55, 15 February 2009 (UTC) | ||
+ | |||
+ | (Edit conflict) You can post your code to subpage, or change to Eclipse. In eclipse, we can config it in project properties, I don't know with Netbeans since I never use it. | ||
+ | |||
+ | Another thing to note for your micros, the best combination of movement is to start with StopNGo, if this fail, move with flat movement. I strongly recommend this movement unless you want to invent new thing to robocode! In order to create good random movement, you must know about: | ||
+ | *Orbital Movement: You should know this because it's basic for many advanced movement. | ||
+ | *Way you move: Are you going to hit wall? Are you stuck in corner/wall? | ||
+ | *Flat Movement: Probably most useful. You will dodge much using flat movement. | ||
+ | I think you don't know the last thing. My first random movement robot that I create without knowing anything above is NanoKitty which result badly. I randomly chose angle and distance, then move. | ||
+ | |||
+ | I think if your continue working on Quartz without reading anything on this wiki, its movement will result in like a basic, unsegmented WaveSurfing for sure. Better if you learn it first. Read more and you will not stuck on your ideas while other has finished implementing it. I've think for rating Virtual Gun with waves for a while before I saw that someone has implements it already with the same formula as mine. Hope you will see the point. | ||
+ | |||
+ | Here are my think of interest of robocode community in nearly future: | ||
+ | * MinimumRiskMovement: it has been invented for a long time but still no one seem to use it in other than melee movement. I saw some robot implement orbit movement as minimum risk. But, still, it can do in many another way! | ||
+ | * Melee/Team stratgy: OneOnOne is explored enough, for now, I think the melee/team strategy is more important especially with team. Currently, almost every teams in teamrumble is just combination of 5 good melee bot. | ||
+ | * Melee Surfing: I don't sure about this, but at least for me. | ||
+ | * Or, if you still want to do in one-on-one, there're 2 interests left: make very good nanos or beat Shadow! | ||
+ | |||
+ | I saw that there 8 people around here. I don't see any top bot release any more version since Oct'08. I don't know if you know more than me since around summer'08, the community isn't as lonely as now. For my list of active robocoder, see [[New Years]] page. | ||
+ | * Me ([[Nat]]) | ||
+ | * [[Starrynte]] | ||
+ | * [[GrubbmGait]] | ||
+ | * [[Rednaxela]] | ||
+ | * [[Voidious]] | ||
+ | * [[Darkcanuck]] | ||
+ | * [[Skilgannon]] | ||
+ | * [[lestofante]] | ||
+ | |||
+ | PS. I'm writing a page long for your little question :) | ||
+ | PS2. I've writing this so long that you finished your first comment ''before'' I finish my comment. The system told me about edit conflict! | ||
+ | » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 18:16, 15 February 2009 (UTC) | ||
+ | ---- | ||
+ | I just don't understand. Why do you add the bot category to those robots? According to Voidious, the page that belong in sub-categories should not be in any of the super-categories of those sub-categories. If you are really adding Bot category, why do you add only some pages? I see that [[WaveRammer]] doesn't get into Bot category still. » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 14:07, 13 June 2009 (UTC) | ||
+ | |||
+ | As I was searching a bot in the bots category, I realised that some of them are not in it but they are in a subcategory of it. I thought that if the majority of bots are also in the main category then each of them should be there. It would also be easier to find a bot or make sure that it exists on the wiki. Anyway, I didn't add the bot category to WaveRammer, I didn't realise that it hadn't got. I only deleted it from NanoBots category as it's not a nano. --[[User:Robar|HUNRobar]] 16:06, 13 June 2009 (UTC) | ||
+ | |||
+ | == Your Robots == | ||
+ | <pre>This is my PM bot designed against other top-bots. Except Ocnirp and MobyNano it beats all of them. :) Anyway, it 20th. ;) </pre> | ||
+ | What? It lost to a lot of robots! See [http://darkcanuck.net/rumble/RatingsDetails?game=nanorumble&name=robar.nano.Prestige%201.0 this page] sort the table by score and you will see :) Anyway, you won't hit the random movement with PM well :) » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 18:20, 27 February 2009 (UTC) | ||
+ | |||
+ | One more, the directly use of StopNGo without flattener are extremely vulnerable to those PM around! I mean, you Vespa lost to a lot of PM robots. » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 18:23, 27 February 2009 (UTC) | ||
+ | |||
+ | I don't know what are you talking about. Prestige lose against only 2 bots from the top 20: Ocnirp and MobyNano. About Vespa: okay, but for some reason Vespa is 9th while your flattenerbot Ocnirp is only 13th. And if you take a look at Pugio, which is 5th, you see that it's even more specialized. ;) | ||
+ | |||
+ | I mean from a whole nanorumble! Note that Ocnirp isn't flattener! The StopNGo is effective way to dodge HOT/LT/CT but not PM nor GF, where RandomMovement dodge PM, but not GF and Flat will dodge GF, and sometimes PM, too. But wait, I'm releasing OcnirpSNG which feature StopNGo + its old random movement! Hope it will be better... » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 18:54, 27 February 2009 (UTC) | ||
+ | |||
+ | I would just like to take this moment to lament that your various bots are kicking the pooh out of my latest bot. Multi-mode, flat(er) movement means one of them is beating me 80 to 20. Yikes. I'll still release it probably tomorrow night though. I've got one more byte to save... :) --[[User:Miked0801|Miked0801]] 06:18, 14 June 2009 (UTC) | ||
+ | |||
+ | I gladly hear about that. :D (both your new bot and the relation to one of my bots.) Yes, I have many bots, but almost each of them are totally different. I like experimenting with them a lot. :) I'll release a new version of Pugio with a velocity PM gun. I use it in my micros and I wonder how it will act in the nanorumble. ;) | ||
+ | :(edit) And I guess the bot of mine you wrote about is probably BlackWidow. ;)--[[User:Robar|HUNRobar]] 14:48, 14 June 2009 (UTC) | ||
+ | |||
+ | Acutally, it's that other weird name bot - but some info for you. Both Pugio 1.47 and MosquitoPM are throwing string exceptions like crazy during the first round of battle (in RoboResearch which shuts down the thread - lame). That's an auto-kill that you can use to improve their score. --[[User:Miked0801|Miked0801]] 05:33, 16 June 2009 (UTC) | ||
+ | |||
+ | Uh.. I can't really understand what you mean. Is there a bug which holds back the performance of my robots? I have never seen any exceptions in their consoles while testing. --[[User:Robar|HUNRobar]] 15:56, 16 June 2009 (UTC) | ||
+ | |||
+ | : When starting a fight in the 1st round (won't happen after) I believe if the two robots start very far apart, your string matching gun throws an out of bounds string exception - usually around index 42 or so. Perhaps preloading your string table with more 0s - though I haven't looked at your code to save for sure if that would help. --[[User:Miked0801|Miked0801]] 16:47, 16 June 2009 (UTC) | ||
+ | |||
+ | : Yes, this could be a problem as my bots have only 30 chars preloaded. Do you think a hotfix would improve the rankings? --[[User:Robar|HUNRobar]] 17:46, 16 June 2009 (UTC) | ||
+ | |||
+ | : Just so you know, I found the same problem with pugio. That might help it's rankings a bit if you fixed that--[[User:CrazyBassoonist|CrazyBassoonist]] 00:50, 29 July 2009 (UTC) | ||
+ | |||
+ | : Ookay, now I had a little time on Robocode, I fixed the bug in Pugio and optimized it a little bit.. Watch out for results, I hope I can beat all of the Neophytes at last. :D --[[User:Robar|HUNRobar]] 14:33, 30 July 2009 (UTC) | ||
+ | |||
+ | :: Just a decent random movement and Simonton's PM gun can already beats all Neophytes. » <span style="font-size:0.9em;color:darkgreen;">[[User:Nat|Nat]] | [[User_talk:Nat|Talk]]</span> » 15:09, 30 July 2009 (UTC) | ||
+ | |||
+ | :: I meant to beat it in the nanorumble. ;) --[[User:Robar|HUNRobar]] 15:17, 30 July 2009 (UTC) | ||
+ | |||
+ | == Questions about flat movement == | ||
+ | |||
+ | Hi all! Now I have a little time for Robocode, and I'm trying to understand the flat and the more sophisticated random movements. I couldn't find and valueble information about flat movement or flattener but I see that most of the complexer bots use this/these. I believe I know what flat movement is (where the graph of the movements visit count stats is flat) but I don't really understand the codes for example in GFTargetingBot. You know, the reverse tuner and those kind of stuff. Can you help me out? --[[User:Robar|HUNRobar]] 16:02, 15 July 2009 (UTC) | ||
+ | |||
+ | : Coming up with a really flat [[Random Movement]] is just dificult and requires a lot of trial and error. [[User:Rednaxela|Rednaxela]]'s [[Carpet/CarpetFinder|Carpet Finder]] is one of the cooler utilities I've seen for studying this. There are different ways to be random and I've certainly put a lot of thought into it, but yeah, trial and error. =) | ||
+ | : Wave Surfers do it differently. Since a surfer is using precise prediction and has complete view of its movement profile, it can simply remember what GFs it's visited and make sure to move to the less visited ones. Of course, this is a lot more complicated than randomly deciding to change direction. And if you're putting in all that work to do correct Wave Surfing, the bullet dodging ends up a lot more valuable than the flattener against most bots. | ||
+ | : --[[User:Voidious|Voidious]] 17:14, 15 July 2009 (UTC) |
Latest revision as of 16:17, 30 July 2009
Please write about Adaptive Oscillator Movement and your new micro bot, please? I'd like to know your ideas. Also, do you want to race with me? I'm 13 years old. Do you want to have some race to top nano or the race into The2000Club? I'm not nano specialist but you are, let's try! » Nat | Talk » 02:01, 15 February 2009 (UTC)
Ook, I'm on it. ;) AOM is strongly under construction, but one of my new nanos will use a more sophisticated version of it. Currently Pugio moves somehow like that. It's about changing the frequency of oscillation in order to find a good solution against hostile guns. However, I've put off AOM recently, because I'm working on my new nano's movement. You mentioned my new microbot. Actually, I have two, but they are not so good at the moment, because I'm too lazy to learn GF. They use almost the same gun that is in my nano's! The difference is that the micros match velocity instead of lateral velocity. This makes their gun inaccurate, but it doesn't get mad by their random movement at least... --HUNRobar 17:00, 15 February 2009 (UTC)
GF seem to be hard, but it heart is very simple! At first, I'm too lazy to learn it. But once I've read more on robocode bot/ranking. The only way to get in top10 of general roborumble is to learn GF or learn TronsGun with DynamicClustering. I prefer first choise at that moment. (Now, I understand both) Once I've got the heart of GF, no problem. I take 2 days understand heart of GF and 2 more weeks to understand how to use rolling average with GF! Once a while, I've clean up the GuessFactor Targeting Tutorial from the old wiki, you can start on that! Feel free to ask me or someone on this wiki for your question.
Another thing, you should go to the old wiki often. There are much more rich information to acquire on the old wiki. One more: Could make your nanos open source, please. I know I can decompile it but almost every nanos release their source code! The robocoder here will give the credit, of course... Ah I think you use IDE, right? is your IDE split up bin and source folder? I think I saw your problem. Now, I've same source and bin directory in order to package robot with source. You may done this, too. Or can can post your code under suppage of your bot page. OR you want them close source? That bad for beginners! (I think I'm enter robocode community later than you, but, sorry, I think I know more theory than you.)
» Nat | Talk » 17:37, 15 February 2009 (UTC)
Don't know, I always tick the include source box. It doesn't include source for some reason recently. I looked up and yes, I've set .../build/classes/ to load my robots from. It means only class files are seen by Robocode. But I don't know how to make Netbeans generate the source into the build folder. --HUNRobar 17:55, 15 February 2009 (UTC)
(Edit conflict) You can post your code to subpage, or change to Eclipse. In eclipse, we can config it in project properties, I don't know with Netbeans since I never use it.
Another thing to note for your micros, the best combination of movement is to start with StopNGo, if this fail, move with flat movement. I strongly recommend this movement unless you want to invent new thing to robocode! In order to create good random movement, you must know about:
- Orbital Movement: You should know this because it's basic for many advanced movement.
- Way you move: Are you going to hit wall? Are you stuck in corner/wall?
- Flat Movement: Probably most useful. You will dodge much using flat movement.
I think you don't know the last thing. My first random movement robot that I create without knowing anything above is NanoKitty which result badly. I randomly chose angle and distance, then move.
I think if your continue working on Quartz without reading anything on this wiki, its movement will result in like a basic, unsegmented WaveSurfing for sure. Better if you learn it first. Read more and you will not stuck on your ideas while other has finished implementing it. I've think for rating Virtual Gun with waves for a while before I saw that someone has implements it already with the same formula as mine. Hope you will see the point.
Here are my think of interest of robocode community in nearly future:
- MinimumRiskMovement: it has been invented for a long time but still no one seem to use it in other than melee movement. I saw some robot implement orbit movement as minimum risk. But, still, it can do in many another way!
- Melee/Team stratgy: OneOnOne is explored enough, for now, I think the melee/team strategy is more important especially with team. Currently, almost every teams in teamrumble is just combination of 5 good melee bot.
- Melee Surfing: I don't sure about this, but at least for me.
- Or, if you still want to do in one-on-one, there're 2 interests left: make very good nanos or beat Shadow!
I saw that there 8 people around here. I don't see any top bot release any more version since Oct'08. I don't know if you know more than me since around summer'08, the community isn't as lonely as now. For my list of active robocoder, see New Years page.
PS. I'm writing a page long for your little question :) PS2. I've writing this so long that you finished your first comment before I finish my comment. The system told me about edit conflict! » Nat | Talk » 18:16, 15 February 2009 (UTC)
I just don't understand. Why do you add the bot category to those robots? According to Voidious, the page that belong in sub-categories should not be in any of the super-categories of those sub-categories. If you are really adding Bot category, why do you add only some pages? I see that WaveRammer doesn't get into Bot category still. » Nat | Talk » 14:07, 13 June 2009 (UTC)
As I was searching a bot in the bots category, I realised that some of them are not in it but they are in a subcategory of it. I thought that if the majority of bots are also in the main category then each of them should be there. It would also be easier to find a bot or make sure that it exists on the wiki. Anyway, I didn't add the bot category to WaveRammer, I didn't realise that it hadn't got. I only deleted it from NanoBots category as it's not a nano. --HUNRobar 16:06, 13 June 2009 (UTC)
Your Robots
This is my PM bot designed against other top-bots. Except Ocnirp and MobyNano it beats all of them. :) Anyway, it 20th. ;)
What? It lost to a lot of robots! See this page sort the table by score and you will see :) Anyway, you won't hit the random movement with PM well :) » Nat | Talk » 18:20, 27 February 2009 (UTC)
One more, the directly use of StopNGo without flattener are extremely vulnerable to those PM around! I mean, you Vespa lost to a lot of PM robots. » Nat | Talk » 18:23, 27 February 2009 (UTC)
I don't know what are you talking about. Prestige lose against only 2 bots from the top 20: Ocnirp and MobyNano. About Vespa: okay, but for some reason Vespa is 9th while your flattenerbot Ocnirp is only 13th. And if you take a look at Pugio, which is 5th, you see that it's even more specialized. ;)
I mean from a whole nanorumble! Note that Ocnirp isn't flattener! The StopNGo is effective way to dodge HOT/LT/CT but not PM nor GF, where RandomMovement dodge PM, but not GF and Flat will dodge GF, and sometimes PM, too. But wait, I'm releasing OcnirpSNG which feature StopNGo + its old random movement! Hope it will be better... » Nat | Talk » 18:54, 27 February 2009 (UTC)
I would just like to take this moment to lament that your various bots are kicking the pooh out of my latest bot. Multi-mode, flat(er) movement means one of them is beating me 80 to 20. Yikes. I'll still release it probably tomorrow night though. I've got one more byte to save... :) --Miked0801 06:18, 14 June 2009 (UTC)
I gladly hear about that. :D (both your new bot and the relation to one of my bots.) Yes, I have many bots, but almost each of them are totally different. I like experimenting with them a lot. :) I'll release a new version of Pugio with a velocity PM gun. I use it in my micros and I wonder how it will act in the nanorumble. ;)
- (edit) And I guess the bot of mine you wrote about is probably BlackWidow. ;)--HUNRobar 14:48, 14 June 2009 (UTC)
Acutally, it's that other weird name bot - but some info for you. Both Pugio 1.47 and MosquitoPM are throwing string exceptions like crazy during the first round of battle (in RoboResearch which shuts down the thread - lame). That's an auto-kill that you can use to improve their score. --Miked0801 05:33, 16 June 2009 (UTC)
Uh.. I can't really understand what you mean. Is there a bug which holds back the performance of my robots? I have never seen any exceptions in their consoles while testing. --HUNRobar 15:56, 16 June 2009 (UTC)
- When starting a fight in the 1st round (won't happen after) I believe if the two robots start very far apart, your string matching gun throws an out of bounds string exception - usually around index 42 or so. Perhaps preloading your string table with more 0s - though I haven't looked at your code to save for sure if that would help. --Miked0801 16:47, 16 June 2009 (UTC)
- Yes, this could be a problem as my bots have only 30 chars preloaded. Do you think a hotfix would improve the rankings? --HUNRobar 17:46, 16 June 2009 (UTC)
- Just so you know, I found the same problem with pugio. That might help it's rankings a bit if you fixed that--CrazyBassoonist 00:50, 29 July 2009 (UTC)
- Ookay, now I had a little time on Robocode, I fixed the bug in Pugio and optimized it a little bit.. Watch out for results, I hope I can beat all of the Neophytes at last. :D --HUNRobar 14:33, 30 July 2009 (UTC)
- I meant to beat it in the nanorumble. ;) --HUNRobar 15:17, 30 July 2009 (UTC)
Questions about flat movement
Hi all! Now I have a little time for Robocode, and I'm trying to understand the flat and the more sophisticated random movements. I couldn't find and valueble information about flat movement or flattener but I see that most of the complexer bots use this/these. I believe I know what flat movement is (where the graph of the movements visit count stats is flat) but I don't really understand the codes for example in GFTargetingBot. You know, the reverse tuner and those kind of stuff. Can you help me out? --HUNRobar 16:02, 15 July 2009 (UTC)
- Coming up with a really flat Random Movement is just dificult and requires a lot of trial and error. Rednaxela's Carpet Finder is one of the cooler utilities I've seen for studying this. There are different ways to be random and I've certainly put a lot of thought into it, but yeah, trial and error. =)
- Wave Surfers do it differently. Since a surfer is using precise prediction and has complete view of its movement profile, it can simply remember what GFs it's visited and make sure to move to the less visited ones. Of course, this is a lot more complicated than randomly deciding to change direction. And if you're putting in all that work to do correct Wave Surfing, the bullet dodging ends up a lot more valuable than the flattener against most bots.
- --Voidious 17:14, 15 July 2009 (UTC)