Bots: FirstBorn and TheOther, currently identical twins
What's special about it?
It is a twin especially (and only working) for the TwinDuel
Tournament, and has a (quite simple) teamstrategy .
Great, I want to try it. Where can I download it?
How competitive is it?
It is in the subtop, dangling around fifth place
How does it move?
Double oscillator with the urge to fight close. Against some opponents it acts like ProvocativeMovement
, as they are forced to flee in a predictable way.
How does it fire?
As of version 0.4 a GF-gun (segmented on distance, acceleration and heading) is used. It will fire at the closest with a preference for the enemy teamleader and weak prey.
Earlier versions used my standard iterative circular gun, known from almost all my other bots.
Does it have a specific TwinDuel strategy?
Although the bots focus their radar on different opponents, their movement is focussed on the enemy teamleader. Therefor it will close in on and try to corner the teamleader. Trying to avoid friendly fire is done by using a 'simulated' radarbeam.
Same strategy goes when one or more bots are killed (2-1, 1-2, 1-1). Alas the second enemy is not noticed anymore when fighting against a majority.
Does it use messaging and what is the content?
The current versions only send own position and enemy position. The earlier versions also exchanged commands like 'move towards bot x', 'lock radar on bot x' and 'target bot x'. However, that was to complicated and errorprone.
What does it save between rounds and matches?
Where did you get the name?
It is a sort of growling to frighten the opponents and it starts with 'Gr' like all my bots. As they are an identical twin their name is a palindrome.
Can I use your code?
It's in the .jar, released under the RWPCL
What's next for your robot?
Tweaking the gun, some more tweaking to the movement and ofcourse some codesize optimizations.
As it now (0.41) seems to consistantly beat Coyote
, the next candidates are GeminiTeam
What other robot(s) is it based on?
It is based on Gruwel
with some GrubbmGroup
(messaging) and Grinnik
as GF-gun source.
- 0.41 - 20071101 (wins 10 - 4 losses)
- use simulated radarbeam iso real radarbeam to detect friendly fire situations
- keep a bit more distance between brothers
- bugfix in firing when gun not aimed
- 0.4 - 20071005 (wins 8 - 6 losses)
- replaced CT-gun by GF-gun, segmented on acceleration, distance and (specific TwinDuel) heading
- tweak to movement to be less predictable
- 0.32 - 20070508 (wins 8 - 6 losses)
- simplified radartarget and movetarget selection (preparation for GF-gun)
- try to reduce friendly fire
- 0.31 - 20070226 (wins 8 - 6 losses)
- much more agressive distancing (close range)
- bugfix startup enemy recognition
- 0.3 - 20070221 (wins 4 - 10 losses)
- fixing radar- and move-Target selection
- back to Gruwel-like oscillator
- 0.2 - 20061102 (wins 1 - 13 losses)
- added communication
- radarlock on different targets (including bug)
- lousy movement
- not very good strategies
- 0.1 - 20060817 (wins 5 - 5 losses)
- basically GruwelTwins with a preference for close-range and a small gun-tweak
Comments, questions, feedback:
I am now busy with a GF-gun for GrauwuarG, but it performs similar to the existing CircularTargeting GrauwuarG 0.32. The gun is taken from Grinnik with the addition of segmenting on heading to make use of the fact that opponents are driven into a corner. I expected a bit more performance than 7 wins, 7 losses and a percentage of 52%. Maybe I will put in the gun of Komarious (is RaikoMicro) as a test to see under what circumstances I do something wrong, as I cannot find any bug yet. -- GrubbmGait
Yeah, I've had a lot of trouble really finding success with improved targeting in the TwinDuel. I tend to think you want lower segmentation than Raiko / Komarious, since you have a lot less data; but it is also a longer match, I guess. Consider how well Shiz does with just HeadOnTargeting. Somehow, I still feel better having a GF gun in there... -- Voidious
I have fitted in the gun of KomariousTeam and it scores 8-6 with 52% . . . . doesn't that sound familiar? This means that my movement is the weak point and nothing else. As one of the few teams with a real team-strategy (even if it is a simple one) I don't want to overhaul it completely, but it is clear that the movement has to be adapted to the close-range strategy. And as the movement is basically the same as Grinnik, Gruwel and GrubbmGrb, I may win some points there too. -- GrubbmGrb
With some minor bugfixes and changes I managed to pump-up my percentage from 59% to 66% with v0.41. It's time to retest some options I discarded earlier.