Leagues

[Elite]

I've been thinking about leagues ...

Griz had an idea a while back about a 'league' where bots had to complete challeges like running mazes. Wouldn't it be so cool to build 'mazebots'. Maybe when walls are working ...

I had a thought recently - how about a 'pond league', where buoyancy could be used. It would allow for some interesting tactics and strategies.
Top-bottom wrap would be disabled.
Maybe PY's floating, pond veg Alga Stratificus could be used in this league, either in addition to Alga Minimalis or replacing it.

How about a 'flying', 'gravity league'?

The 'gravity league' may use pond mode too, so bots could float against gravity using buoyancy settings.

Another thought I had was a 'team league'. Two 'bot teams' compete in an empty arena with no vegs and no costs

How about a 'king of the hill' league, with one massive, sterile veg in the center of the arena and the other bots fighting to control it.
Of course, no interfering with the veg (changing it's memory values)
A fun variant of this game might be to have the veg slowly rotate and spray out ultra-powerful shots.

Thoughts? Other ideas?

[Henk]

I really like the wall/maze idea  How about a league in which bots have to find their way to the center of the sim? (through a maze of course) then note the time it took to get there (in cycles) as the time to compete with in the league.

It would require a complete overhaul of the wall system though I think.

[Numsgil]

If I ever actually complete the env grid, I wanted to implement "dirt".  Bots could remove and redeposit dirt elsewhere, creating (I hoped) some realistic ant nests.  I think that would be cool.

[Endy]

Alright is anyone else having walls work for them in 2.37.6?

They've been working fine in mine and the add-ons I've made

Mainly I've been doing enviroment splitting simulations, also some cornucopia type scenarios with the food falling down from it's "pot" and feeding the animals below.

[Henk]

They've been working fine in mine and the add-ons I've made

They have in mine, too. My biggest concern is that they are currently quite a load for the CPU. Building big mazes slowes the program down a lot.

(I did a quick test, on a random sim I got 25 cycles/sec without walls and 7 cycles/sec with a maze. Sidenote: I have a 900mhz cpu and the version I ran is Endy's)

[Elite]

The entire concept of walls in DB seems flawed. Walls are just a special type of bot. Walls should just be opaque lines that bots can see and colide with
There should probably also be a new sysvar: *.wall
A maze-generation option would be nice too

How hard would it be to implement in 2.4?

******

How about a 'minotaur league' - bots fight each other in a maze

[EricL]

How hard would it be to implement in 2.4?

It would be...work.  Not hard, just a significant number of lines of code.  The reason walls are implemented as bots today is so that all the existing collision detection, etc. can be leveraged.  As you correctly point out, this is seriously flawed and is the reason walls slow things down so much.  Properly implmented, walls would have a minimal impact on execution speed but new code for collision detection and perhaps ways for bots to detect walls would need to be written.  I don't think we need any new sysvars for building walls themselves, but there are open design questions that may.  Should bots be able to see differeces between walls and bots for example?  That would require tweaks at least to the bot vision system, which I think is what you are suggesting with the *.wall sysvar.

Adding maze drawing once cheap walls are in place would not be difficult.

In my opinion, the degree to which wholly new features/concepts such as this get implemented in 2.4 is completely a function of how far away the C++ version is.   The more we add to 2.4, the more that needs to get re-implemented in 2.5+.  That said, I think walls would add a lot to both evo and combat sims, providing a tremendous leap in environmental complexity and facilitating such things as population isolation, opportunities for speciation, genetic drift, etc.

[Numsgil]

The physics engine in 2.4 isn't 100% built in such a way that walls would be easy.  Walls would be much easier to implement in 2.5.
 
 Speaking of which, I'm getting closer   Or I sure hope so.

[Elite]

I was looking back over some of the posts in the old old forum, and an interesting idea there was to have a league where you were not allowed to use any of the movement commands (.up .dn .dx and .sx)

Contenders elegable for this league would include:
 - PY's Inchworm series
 - My comet-worms and rocket-bots
 - Abyaly's Turbo

Interesting idea, don't you think?

[Numsgil]

It woul;d be pretty easy to set up, all you'd need to do is turn down the "movement efficiency" in the physics options to disallow movement types that you don't want.

[abyaly]

Turbo uses the move commands.
But right now Im playing with a MB that wiggles to move ^^