Robo Home | Changes | Preferences | AllPages

Peek at Pearl's code Help a math challenged geek out

Bulletin Board

25 May, 2004 - Refactored the gun and useing Raiko's reverse calcuations while I figure out a better way to do it myself.

16 May, 2004 - Nothing special or different. Some code clean up and a some new segments in the gun. The wall segment still is not performing as well as I would like but I left it in for now.

16 February, 2004 - Interim release. Sqeezed a lot of bytes out of it and it has some know issues that need to be resolved but I wanted some feedback on the direction I am taking.

08 February, 2004 - Released BP .82. Who knows anymore. This version has a pretty good movement according to it's own gun so I am sure it will plummet even further in the rankings. Once again it is capable of beating some very good bots while losing miserably to others. I also updated BPs code page so have a peak if you are curious.

27 January, 2004 - Released BP ver .81. I mistakenly released a version .80 which had some data left over from testing. This version was getting interesting results in testing beating some very impressive bots while loosing to some equally un-impressive bots. Wall smoothing is in again. I still suffer from turning too far sometimes coming off the wall but I am not sure how to fix it. If you want to help (and you know you do) have a peek BP's code and see if you can figure it out.

24 January, 2004 - Released BP ver .79. I am playing with wall smoothing parameters to see if I can correct some movement issues. I have updated the code page with the correct code as well. Anyone who takes the time to run and veiw the code care to explain and help me solve the circle off the wall issue? I know it has something to do with turning the shortest distance but I can not seem to figure it out. -- jim

19 January, 2004 - Release BP ver .77. Features the return of wall smoothing. I removed any Musashi style tricks and the VertiLeach killer segment. I have also switched the distance index to segment on bullet flight time. Some other small changes but nothing major. Seems to be even more so vulnerable at zero. -- jim

13 January, 2004 - Relase BP ver.75. Internal code change and a new VertiLeach killer segment =^>. Can you guess what it is PEZ? -- jim

10 January, 2004 - Release BP ver .73. Mostly a roll back to ver .69. Experiment failed. Time to take another tack. -- jim

6 January, 2004 - That was weird to type for the first time this new year! Released BP ver .71. This is a highly experimantal version of BlackPearl featuring some never before tried features. I have implemented a wall smoothing algorithm that I hope will make me more effective vs some of the better bots segmenting on proximity to that boundary. I have finally added my own segment concerning wall segmentation too. Finally I have replaced my directional timer with a random engine that checks each tick for a reverse probability. All in all, this release should not be too much worse than the last version but it may be =^> Still vulnerable to head on targeting and I need to implement a solution to that at some point but I want to first see how my wall smoothing algorithm works. I will post the code on the code page in a few days if I decide to keep this version.

28 December, 2003 - Current release is .67. I found and fixed a subtle movement bug that was affecting near wall performance. The bug made Pearl "yo-yo" against the wall more than it should have. This bug is also present in all versions of Jekyl so I will eventually need to correct that too. I have added some very primitive melee capability although I must say it is very horrible. Code size has been shrunk some too as I am trying to find about 100 bytes to try some wall smoothing. If I can not find the bytes else where I may just remove the data saving stuff and make like ABC's bots =^>.

11 December, 2003 - Released .62. Features a movement tweak that I think will help some. Code size is actually smaller.

7 December, 2003 - .60 is out. I officially give up on Pattern Matching as I can not bloody well seem to be able to accomplish it with any sort of confidence. You can forever more think of BlackPearl as mini Jekyl from here on out.

3 December, 2003 - .54 officially stunk. Note to self: more testing would be nice. Released .55 which is a move away from my super spiffy reduced wall and corner avoidance which apparently did not work all that well.

2 December, 2003 - I am officially bored with my other bot now so I have released a new version or Pearl (.54). Pearl has languished for too long. For this release I have refactored the pattern matching gun and finally got around to making an attempt at solving the MultipleChoice problem. I have not taken any extreem measure like others to tackle this though. I guess you could say I ran home to mom as I have converted all of my potential aiming solutions to a faux GuessFactor and then choose the factor with the highest occurance. Seems better than before but I am sure it is not that great. Still no variable length pattern matching but I have some 100 bytes left so maybe I can wedge it onto that.

1 Sept, 2003 - Pearl makes her debut on the Eternal Rumble. Note that this bot was not developed without a single iota of melee ability. I am pleased with the #59 spot as I spent about 2 days working on this bot.

Currently on my todo list:

Currently working on:

What's special about it?

It is my first mini. It will be my first open source bot.

Great, I want to try it. Where can I download it?


How competitive is it?

Seems to do fairly well. Movement is not all that sound but it is not to terrible.

How does it move?

Scaled down TargetOrbit? System taken from Jekyl.

How does it fire?

Uses a guess factored targeting system copied from Jekyl.

How does it dodge bullets?

Changes directions when it is fired at.

How does the melee strategy differ from one-on-one strategy?

one-on-one only at this point. Probably always will be.

How does it select a target to attack/avoid in melee?

It doesen't do melee if it can help it.

What does it save between rounds and matches?

It saves everything it can as often as it can.

Where did you get the name?

I have always had a fascination with pirates. I saw Disneys new movie The Pirates of the Carribean. You do the math =^>

Can I use your code?

When it's released help yourself. I only ask that if you improve it please post your changes here.

What's next for your robot?

Improve the movement.

Does it have any WhiteWhales?

None right now. More to come later I am sure.

What other robot(s) is it based on?

It is based on a lot of ides discussed here and in other locations.

Comments, questions, feedback:

I'm glad you gave up on the PM gun. It got VertiLeach to lose the #1 spot in the mini rankings. Now I think it will take a day or two before Verti is again mini-king. =) -- PEZ

Yeah, scrap data saving. Let's see some wall smoothing instead. Good movement pays more than saving data in this competition format anyways. -- PEZ

Go for melee. I always think it's good when my competitors concentrate on several things at once. =) -- PEZ

I am beginning to wonder if you are sincere. Is that fear I smell? =^> -- jim?

I just think it's good that melee and stuff is there so that the creativity of my competitors can be scattered a bit. What I would do if I were BP's creator would be to make one 1v1 and one melee version of it. But please don't do that. =) -- PEZ

Thanks for pointing out that weakness in Verti. Now I can make it stronger against these new gen bots. =) -- PEZ
Btw, if you want to shrink the codesize a little, you can use Rules.getBulletSpeed?() instead of using your own. --Starrynte

Although it could be done, I think that it would not be a good idea because many of the clients are using older versions without support for the rules class. -- Kinsen

Robo Home | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited December 28, 2006 20:46 EST by Kinsen (diff)