INFO   Main   Docs   FAQ   News   Tutorials   Interviews   Forums   Mailing List
ONLINE    BATTLE ARENA   Accounts   Uploads   Downloads   Hall of Fame
|   No-Limits Challenge: JROBOTS FOREVER! (January 6th 2010)   ||   October-December Challenge: The End! (December 26th 2009)   ||   October-December Challenge: JROBOTS 2K9 Q4! (October 3rd 2009)   ||   September-October Challenge: JROBOTS 2K9 Q3! (July 4th 2009)   ||   April-June Challenge: JROBOTS 2K9 Q2! (March 28th 2009)   ||   Old News  |




Hall of Fame





Mailing List
Contact leo
In Defence of the Original Robots

(Excerpt from the mailing-list)
About the death of Shark Leonardo wrote:
> I don't want to upload Shark any more: it is my weakest robot and
> I've lost its source code in a crash of my HD, so I can't improve it :(

I liked Shark very much. Sure, it wasn't exactly strong, but it was so untipical.
To be true, I'm not very happy when I see people uploading Rabbit or Rook, and I don't like either when they do a Platoon-clone based on the tutorial.

Many people put a lot of efforts into the coding aspect of this game, instead of thinking about new strategies; me and BingleFish are not professional programmers, but we are both engineers, and we think we've proven that analysis, physics and a good strategy can yield top-notch results, I see many of you begging for API calls and OOP to improve their robots, but I want to say, there's no need for fancy data structures, esoteric objects, clean interfaces; all you need is: arrays; physics; ideas.

Some people needed explanation about Leonardo's enemy speed calculation functions, I obtained it by myself a lot before it was published on the tutorial, and there's so much more under KB's hood that many of you would be surprised to see it if we decide to unveil it; in my vision this is a game for engineers, not programmers, if you don't have the necessary knowledge of math, physics, and even signal processing theory is useful, then you've better turning your attention toward other competitions, there're so many sheer programming championships that you won't miss JRobots.

I have to say that even Leonardo somewhat encouraged this approach ("don't innovate, tweak and refine instead") because since the beginning he clearly explained his robot's strategies to the others, favouring clone-makers; if you lose the strategic advantage, you can only win by having more efficient code, but that's a lot affected by the well known cpu-speed issue, and in the end, a better idea is a lot more effective than weeks of tuning, IMHO, even if ideas are usually easier to copy than code.

When I developed my first robot, I didn't even take a glance at the tutorial, in fact I believe that Monty (the original) was the first competitor to leave the Platoon-like fashion (run straight at top speed) to try something new (non linear movements); the older players will surely remember the 3 strongest robots of the first challenge, Chaky, MarukoV4 and Vampire, they were improvements over Platoon but they were very similar to it in behaviour; then it came (I believe) Jimbo with the first circular movement, then MontyZ and Tiny with a zig-zag path.

Shark's circling at corners (like cartoon's sharks) is so nice that I'd hate not seeing it in action anymore; when you look at a team match which involves for ex. Shark, Fish, KillerBees and Phalanx, you find it a lot more entertaining than a match where all the robots cross the arena more or less erratically: you see the sharks haunting the corners, with their slow and menacing circling; the phalanx advancing tight and steady like a real army; the bees buzzing around their nest; the fishes roaming left and right like in an aquarium...

Two other robots I like are Pulse1 and Epa1, they're both very innovative and show uncommon features, I don't want to discuss them in depth because I don't want to harm them by exposing some of their strategies to the public; Marvin3 is cool, but I find its pattern movement too similar to Epa1's one to call it innovative, and it has still to prove it's stronger than Epa even in double play.

Me and BingleFish could possibly decide to publish KillerBees source code, but I'm wary of doing so since it could mean a wave of new KB's clones...

Walter (author of KillerBees)

To join the mailing-list visit OneList
For more information
send an email to
Project Hosted by SourceForge Copyright © 1999-2009 Leonardo Boselli
All Rights Reserved. Legal Terms.