Dookious
Jump to navigation
Jump to search
- Sub-pages:
- /Version History - /DookiCape - /DookiLightning - /DookiSaberRRGC
- /Archived Talk 20060413 - /Archived Talk 20061129 - /Archived Talk 20071111
| Dookious | |
| Author(s) | Voidious |
| Extends | AdvancedRobot |
| Targeting | GuessFactor Targeting |
| Movement | Wave Surfing |
| Best Rating | RoboRumble: 2131 (1st) |
| Current Version | 1.58 |
| Code License | RWPCL |
| Download | |
Background Information
- What's special about it?
|
- Great, I want to try it. Where can I download it?
| Current version: | 1.58 | |
| Best version: | 1.573 |
- How competitive is it?
- It's very competitive. It's currently the #1 rated bot in RoboRumble General 1v1 (since 4/18/2006) and it trades with Shadow and Phoenix for the Premier League throne.
Strategy
- How does it move?
- It uses Wave Surfing. See the DookiCape sub-page for more details.
- How does it fire?
- It uses GuessFactor Targeting with two Virtual Guns, one tuned to hit adaptive moment. See the DookiLightning sub-page for more details.
- How does the melee strategy differ from one-on-one strategy?
- There is no melee strategy - it's only designed for 1v1.
- What does it save between rounds and matches?
- Between rounds, it saves all the targeting, virtual guns, and Wave Surfing data, plus some general hit statistics (percentages and damages). Between matches, it saves the best GuessFactor bin for all segments that were visited more than a certain amount (i.e., SuperNodes).
Additional Information
- Where did you get the name?
- I named the tank after Count Dooku, since he seems like a duelist in the Star Wars mythos, and adapted that to a style that fits my own handle.
- Can I use your code?
- Yes, feel free! It's released under the RWPCL. The gun and the movement are pluggable, a la CassiusClay, so knock yourself out.
- What's next for your robot?
- Throne defense. =)
- Does it have any WhiteWhales?
- Phoenix has been hot on Dooki's heels on and off for a while. It would be nice to one day beat Shadow and Phoenix consistently, but that day may never come.
Credits
Dookious is not directly based on any other robots; I have, however, learned a TON of information from this site. As far as using other people's code, the only direct code I'm using is Albert's FuturePosition classes, which I am very grateful for.
I also simply *must* say thanks to everyone who has contributed to this site, and to PEZ for hosting it. Robocode just wouldn't be the same experience for me without it. That goes for PEZ, Kawigi, Pulsar, David Alves, Paul Evans, ABC, Axe, Corbos, wcsv, The Martinator, GrubbmGait, Kev, Simonton - ok, everyone that's part of this great community. Hopefully I can give back at least some fraction of what I have taken from this wiki.
Explicit credits need to go to:
- WaveSurfing - invented by ABC
- GuessFactor Targeting - invented by Paul Evans
- FuturePosition - A great piece of code from Albert to predict the result of different movements.
- RollingAverage - A simple, elegant concept from Paul Evans that made its way into Dookious. I also adopted a very minor, but notable, implementation idea from PEZ.
- WikiTargeting - methods for analyzing GuessFactor data invented by Vic Stewart, used in Dooki's data saving.
- CassiusClay - I have learned many, many things from this excellent tank, from general code structure, to very specific segmentation methods for certain types of data (especially wall distance!), to various other small details that go into making a great tank. Thanks, PEZ!
- Ascendant - I'm almost positive that Mue's comment about using a second gun tuned for AdaptiveMovement was my initial inspiration to try my hand at an AntiSurfer gun, so props to him for sharing that insight.
- TheArtOfWar - This is the tank I most remembered from my initial stint in Robocode a few years ago, and it set the bar relatively high in the early days of benchmarking Dookious. The (admittedly cheesy) Count Dooku quotes that Dookious prints to the console on victory is a bit of an homage to this classic tank.
- Corbos and wcsv - I've topped them in rating for now, but these guys provided some much needed, active competition when I was really gaining traction with Dookious. I look forward to vying for rating with them once they catch up =)
- PEZ for Iterative WallSmoothing algorithm, Kawigi for entropy routines - I'm no longer using a version of PEZ's iterative WallSmoothing, but many previous versions did, and I used it as reference to write a faster method that does the same thing. I also made use of some entropy routines that Kawigi posted in my early research into segmenting a GuessFactor gun.
- David_Alves/DrawingBot - Much thanks to David for saving me time developing drawing methods for the new Robocode graphics capabilities! It was very helpful when rewriting Dooki's movement.
