Contest: Animal/Bot Coding (AI)

[t i l e x]

Maybe you should fix rules since for now there is nothing that is considered as cheating except modifying the Animal class

[Corsix]

Maybe you should fix rules since for now there is nothing that is considered as cheating except modifying the Animal class

All will be revealed shortly...

[Corsix]

Contest has begun...
Contest: Animal / Bot Coding (AI)
Aim:
To create the best animal, eg. the one that survives longest in the virtual world.
How it works:
You derive a class from either Herbivore or Carnivore and then use the built in methods to interact with the virtual world.
Submit:
Upto 5 .h files which include your animals and or any helper functions. Please prefix all files,classes and functions wih your name to avoid conflicts.
Get Started Now!
You will need to read the basic guide (here)
If you want to test your creation then you will need to D/L the test environment. There are many version available, the most up to date is 103b.
Build 100: 21k - here
Build 101: 25k - here
Build 102: 65k - here
Build 103b: 25k - here
Build 104 patch (from 103/b): 6k - here
You may also want r4nd0m's win32 GUI for the test environment:
Version 0.2 for build 103b : here
Any questions?
Post 'em here and I will be happy to answer
Deadline
All animals must reach me (Corsix@PlanetUnreal.com) by 5:00 PM GMT, Saturday 4th of September. However, after that the contest will probably be re-held fortnightly (eg. you can improve your bot and then re-enter).
Rules
You cannot overrive ANY members of Animal/Herbivore/Carnivore that are not declared virtual.
You MUST NOT call/access and of the __int__ functions or variables declared in Animal. Also you MUST NOT call __int__EndOfLife, __int__MakeVictor, __int__DistanceBetween or __int__NewAnimal .
No more than 5 (five) animals to be submitted per user per round.
Have fun

Oh and just so that you know, I am using M$ Visual Studio 2005 Beta 1 on Windows XP for compiling.

To use the test environment, include your animal headers near the top of Animal.cpp and create your animals in main() near the bottom.

[Darobat]

Good job Corsix! I havent dld it yet because im lazy! Looks good though. This is gonna be one hell of a match!

[Ryan]

I've run the code a dozen times or so, only twice has anything happened after 1000 turns. Both those times the herbivores started breeding, and the population very quickly grows to the point where turns take several seconds to compute. The carnivores never did a thing and stayed at a population of 10 the entire time.

[Corsix]

That's because my Herbivore class is fairly good - whereas the carnivore isn't very clever. If there are bugs / errors in the test environment then report them and they can be fixed.

PS. Thanks Alvaro for moving this to the right forum

EDIT: It seems that in my rush to post the code I missed a few bugs in the test environment - they will be fixed and the new code posted in an hour or two.

EDIT2: Fixed See the above post for the new test environment.

[Tenji]

darn so i cant make my suicidal furbies that explode when you eat them

[Corsix]

darn so i cant make my suicidal furbies that explode when you eat them

Nice idea maybe in the next round....

[Lovechild]

and so it begines


you will all bow down to BOO

[RecursiveS]

Very nice work Corsix....

I really wish I had the time.....

This could make you money (see open disscussion, selling sites!!!!!!!)

I'm really gonna try and carve some time out for this one, cos I've got a hamster that will kick butt!!!! ...........(in my head at the moment )

[r4nd0m]

Hey Corsix, I tried compiling ur demo-world, and got these errors:

Code: Select all

D:\c++\animalai\AnimalBot\Animal.cpp(41) : error C2562: '__int__EndOfLife' : 'void' function returning a value
        d:\c++\animalai\animalbot\gamedefs.h(71) : see declaration of '__int__EndOfLife'
D:\c++\animalai\AnimalBot\Animal.cpp(50) : error C2562: '__int__MakeVictor' : 'void' function returning a value
        d:\c++\animalai\animalbot\gamedefs.h(72) : see declaration of '__int__MakeVictor'
D:\c++\animalai\AnimalBot\Animal.cpp(477) : error C2065: 'random_shuffle' : undeclared identifier

I'm gonna work on fixing them right now, so I can test it out, just thought I should let you know.

[..edit..]

Also, when reading your AnimalContest.txt, I found conflicting statements.
Near the top, it says if you're energy goes to 0, you die. In the attack section, it says if you're size goes to 0, you die. It also says if you use up all your energy during a fight [so you're energy is 0], the other person keeps attacking till their energy goes to 0. wouldn't that mean you both die? Can I have a bit better explanation on the stats, please? thx!

[..edit2..]

also, in the Tick() function, are we allowed to Move/Eat/Breed only once? For example, someone could just put:

Code: Select all

for(int i = 0; i < 20000; i++)
{
  Eat();
  Move(Corsix_RandDir(), 1);
}
Whick would eat a heck of a lot. Or, the creature could run around the entire world, gobble everything, breed a million times all in one turn..


[Tenji]

i got same errors. i'm a noob so not sure how to patch that so i can get it to compile and stuff.

[r4nd0m]

Oh, also in Animal.h, you might wanna add:

Code: Select all

#ifndef _ANIMAL_H_
#define _ANIMAL_H_

.. code goes here ..

#endif


cuz when I added my own header, it got redefinition errors.

for the error that doesn't recognize random_shuffle or w/e, just stick an std:: in front of it [random_shuffle() is part of the std namespace]. for the other two errors, I made the function types ints, returned 0 at the end to make sure all paths return something, and changed the definitions of the functions. When I run the app w/ Corsix's 'super carnivore' it seems to win in one round. If I comment the super carnivore out, it seems to run forever, and runs so fast that I can't see it. Also, when it finishes, It doesn't print the table of winners [i'm not sure if it's supposed to do that..]

[..edit..]

also, it seems that only the last animal that gets added in main() wins [or at least keeps going,or even does anything, no other animals get reported..]

[Corsix]

Thanks for your support guys. I worked for the best part of three days with no major breaks to get the code to where it is now so its probably buggy...

You raised some valid issues, it compiled fine for me but I will see what happens in Dev-C++ (as most people use that or can get it for free).

[Corsix]

Also, when reading your AnimalContest.txt, I found conflicting statements.
Near the top, it says if you're energy goes to 0, you die. In the attack section, it says if you're size goes to 0, you die. It also says if you use up all your energy during a fight [so you're energy is 0], the other person keeps attacking till their energy goes to 0. wouldn't that mean you both die? Can I have a bit better explanation on the stats, please? thx!

Yes, it is a little unclear. Each animal gets a number of hits equal to half the amount of energy they use (eg. 10 energy = 5 hits). They take it in turn to hit until one animal has ran out of hits. Then the one with hits will hit continuosly until he has also run out. Each hit decreases the opponents size by your power. When size hits 0 they die. If your opponent dies and you still have hits left they are turned back into energy.

also, in the Tick() function, are we allowed to Move/Eat/Breed only once? For example, someone could just put:

Code: Select all

for(int i = 0; i < 20000; i++)
{
  Eat();
  Move(Corsix_RandDir(), 1);
}
Whick would eat a heck of a lot. Or, the creature could run around the entire world, gobble everything, breed a million times all in one turn..


I thought of that and I put (or meant to put) "no infinite or near infinite loops in tick". However a limit can be imposed

The test environment has been tested in:
M$ Visual Studio 2005 Beta 1 - No Problems
M$ Visual Studio .net 2003 - No Problems (thx Roboticguy2)
DEV-C++ 4.9.8 - No Problems
M$ Visual C++ 6 - Issues relating to template functions. See the comments in main() for a fix.

PS. Test Environment has been updated to build 102