Bots and Simulations > DNA - General

Top bots who are still build truely modular?

(1/3) > >>

Welwordion:
Personally I tend to focus on  certain aspects of bots I try to realize, improve or test and have not much interest in other parts (for example I never did much work considering the control of shoottype and strenght).

However I have certain problem to test this cause there seems to be few bots around that are goodly  documented or neatly seperated turning, movement, feeding genes, repro genes etc and such where I could just replace the old genes with the genes I thought up and then let the bots compete with each other.
So as I have not much interest in building all those fighting gimmicks myself, my bots never ever get into bestiary or competition with other bots.

The result is that genes/ideas that might be a potential improvement or an interesting addition to the current ones are never seen by anyone but the creator and can not be tested reliable.
(For example I would like to attach my beholder movement and aiming system onto multiply's metabolism and shooting regulation system to see which one survive)

Originally the gene depository was supposed to be something like that search the genes together you need add your own genes and test it out but see what became of its seem you can not find one normal shoot feeding gene in there.(atleast I could not)

shvarz:
I did not know we had a gene depository. This might be a great resource and useful for newbies. Choose from existing genes, combine and get a bot. Nice!

Numsgil:
Genes aren't quite plug and play like that.  Something complex, especially a bot using SG coding techniques, is going to step on a lot of other genes to operate.  So usually when a gene is posted it's more of a proof-of-concept than a ready solution.  Coders usually treat it as such.

Animal Minimalis is the best documented bot, and is what I would suggest using to test new bots with, because it can be entirely understood without too much effort.  However, even with this simple bot, the 1st and 2nd genes aren't independant.  They have entanglements with each other.

Welwordion:
Ah no, animal minimalis is not really top notch you can not polish or test your bots if you use bots that are under the current standard that has to be overcome.

So I will now do what I rather would like to not do, work trough every single gene of multiply and identify its function. Maybe I will even create some documentation.

On that note

cond
*.val20 !=
*.loc2 0 !=
start
*.val2 *.loc2 store
stop

I think thats wrong ----> *.val20 !=  
Anyone has a current version of multiply to check if that was ever corrected?

cond
*.robage 0 =
start
.tie inc
.shoot .memloc store
.venval -2
.shoot .vloc store
.myeye .ploc store
stop

A store is lacking? ---> .venval -2

Attached my first run over the code, was easier than I thought, oh and yeah I realize I am one whiney complaining bastards.

MacadamiaNuts:
I think future versions could use a GUI bot builder with a 'plugin' genes repository file.

Each plugin would be a txt file in a 'pluginsfolder/functionsubfolder/', the GUI would be like the Species menu, with the window to see the header comments and 'up/down' controls. Then the 'make' button would prompt for a save location, and it would just merge the selected plugins into a bot, comments included.

Navigation

[0] Message Index

[#] Next page