Skirmish AI: E323AI 3.22.4 - Page 9

Skirmish AI: E323AI 3.22.4

Here is where ideas can be collected for the skirmish AI in development

Moderators: hoijui, Moderators

User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Ok, ive been playing with this a fair bit firstly, mega congradulations, this AI kicks ass! beats all the other AIs and performs reasonably vs humans, and the raidering microing is incredibly cool!
thx :)
however, the CPU usage is excessive- considerably higher than any other AI, and enough to lag my machine. Im guessing that is somthing to do with the huge amount of constant threat checks, since units seem to receive about 2 orders a second. might want to tone this down until the system usage is reasonable.
CPU usage in the latest version is very low (look in the log at the bottom "ScopedTimer"), however memory usage is quite high at the moment. This is fixed in git though.
with defences, with both XTA and BA a quick and easy way to implement defences effectively is place an imperative to build one the cheapest defences (xta missle towers/BA llts) within 200 of metal extractors
Agreed, this will be going in the defensive model for the new major release.
AIs also play TA mods far better when they have build in build orders for labs (ie, instead of making random shit mix of units, they follow a sequence from a txt file. To hide this, you could have 5-10 diffrent lab build orders in the file so it varies between them)
This could be a very good idea actually, I'll work it out.
the other thing that would greatly improve playing performance is to implement reclaiming- ie, for metal piles worth over 50, cons can have an action to reclaim them. increase the priority of reclaiming in proportion to the quantity of metal in wrecks and the distance of the wrecks from the con. in BA/XTA, have fark/necro performing purely as reclaimers.
Also on the todo list already :)
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Re: A new artificial intelligent opponent - E323AI

Post by Argh »

[ 0] Player Player (auto)-paused the game for letting Skirmish AI E323AI initialize for controlling team 1. The game is auto-unpaused as soon as the AI is ready.
[ 0] SkirmishAI1: Speedmetal infects my skills... I won't play
[ 0] SkirmishAI1: *** E323AI 2.12.0 - Ultralisk ***
[ 0] SkirmishAI1: *** Error323 - folkerthuizinga@gmail.com ***
[ 0] SkirmishAI1: *** This A.I. mainly focusses on the XTA and BA mods ***
Crashes to desktop.
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Yeah, its not happy on speedmetal maps. Metal searching would be epic slow since there are so many spots. Try non-metalmaps first.
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Re: A new artificial intelligent opponent - E323AI

Post by Argh »

P.U.R.E. is not played on "non-metalmaps". It uses a specific map format. Surely there are ways to keep AIs from going beserk if there's metal everywhere, like, IDK, using the extraction radius to determine the ideal spacing?
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Ah hehe, well yes there are, but not in E323AI. Sorry :/
User avatar
AndiNo
Posts: 24
Joined: 23 Mar 2009, 09:45

Re: A new artificial intelligent opponent - E323AI

Post by AndiNo »

I was told to try your new AI as the other AIs seem to do mostly nothing when playing XTA with fleabowl (I'm using XTA 9.59). But it crashed when I added your AI to the game. Here's my log (I put infolog and the AI log together):
http://pastebin.com/m459e994c
Tested this on DesertTriad and TacticalDividev2.
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Re: A new artificial intelligent opponent - E323AI

Post by Argh »

So, basically, you can't be bothered to fix such a rudimentary problem? Okie doke.
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Hi AndiNo,

There seems to be a problem with the new XTA and E323AI. I'll look into it and try to figure out the problem. The previous version should work fine.
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Argh wrote:So, basically, you can't be bothered to fix such a rudimentary problem? Okie doke.
Well, I would love to but my todolist is growing at an incredible rate atm :(. It would take a while.
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Re: A new artificial intelligent opponent - E323AI

Post by Argh »

No, it wouldn't, compared to most of the feature requests you get :roll:

Anyhow, np, you're just another OTA AI guy. Pretty much all we have here, no harm in it, have fun.
User avatar
AndiNo
Posts: 24
Joined: 23 Mar 2009, 09:45

Re: A new artificial intelligent opponent - E323AI

Post by AndiNo »

Alright, I'll just wait then. :)
User avatar
1v0ry_k1ng
Posts: 4656
Joined: 10 Mar 2006, 10:24

Re: A new artificial intelligent opponent - E323AI

Post by 1v0ry_k1ng »

Argh wrote:P.U.R.E. is not played on "non-metalmaps". It uses a specific map format. Surely there are ways to keep AIs from going beserk if there's metal everywhere, like, IDK, using the extraction radius to determine the ideal spacing?
the pure system is dumb beyond belief. to fix it to the same effect but less annoyance, just place a grid of metal spots on your pure maps, but whatever, I guess you didnt want ta cross compatability.
User avatar
hughperkins
AI Developer
Posts: 836
Joined: 17 Oct 2006, 04:14

Re: A new artificial intelligent opponent - E323AI

Post by hughperkins »

Error323 wrote:The appropriate config file is in git and in the tarball.

And try again :)
Here you go ;-)

http://manageddreams.com/hughaireplays/ ... df.tar.bz2

I'll admit it was close, too close to call really. I feel that if I played multiple times, HughAI might not win as many as half.

But, it won this one, and it was the only one I tried.

Hugh
User avatar
TheMightyOne
Posts: 492
Joined: 26 Feb 2007, 14:32

Re: A new artificial intelligent opponent - E323AI

Post by TheMightyOne »

Argh wrote: Anyhow, np, you're just another OTA AI guy. Pretty much all we have here, no harm in it, have fun.
you're very fast at judging people
User avatar
hoijui
Former Engine Dev
Posts: 4344
Joined: 22 Sep 2007, 09:51

Re: A new artificial intelligent opponent - E323AI

Post by hoijui »

Its Arghs way of trying to get ppl to do stuff for him; basically the same technique that small kids use for bragging.
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Heh, then he should know its often very counter productive =)
bbe
Posts: 10
Joined: 15 Nov 2008, 17:24

Re: A new artificial intelligent opponent - E323AI

Post by bbe »

Hi Error323

I tried your latest master last night, with the CPU improvements (which were very good). I have one problem, however, the AI only seems to build fleas. I've found the log an it's correctly parsed the cfg file... the log is filled with lines like this:

Code: Select all

[01:37] (EE): CUnitTable::canBuild failed to build MOBILE | ASSAULT
[01:37] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ASSAULT
[01:38] (EE): CUnitTable::canBuild failed to build MOBILE | ARTILLERY
[01:38] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ARTILLERY
[01:38] (EE): CUnitTable::canBuild failed to build MOBILE | ARTILLERY
[01:38] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ARTILLERY
[01:38] (EE): CUnitTable::canBuild failed to build MOBILE | ARTILLERY
[01:38] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ARTILLERY
[01:39] (EE): CUnitTable::canBuild failed to build MOBILE | ARTILLERY
[01:39] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ARTILLERY
[01:40] (EE): CUnitTable::canBuild failed to build MOBILE | ARTILLERY
[01:40] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ARTILLERY
[01:40] (EE): CUnitTable::canBuild failed to build MOBILE | SNIPER
[01:40] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | SNIPER
[01:40] (EE): CUnitTable::canBuild failed to build MOBILE | ASSAULT
[01:40] (EE): CWishList::push failed for Kbot Lab(142, 116) categories: MOBILE | ASSAULT
[01:40] (EE): CUnitTable::canBuild failed to build MOBILE | ASSAULT
Any ideas?
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

Hey bbe,

From the looks of it, it used the generated configfile instead of the one provided in git. Note that you have to copy the configs directory manually from the git repo to ~/.spring/AI/Skirmish/E323AI/2.12.0/.

I need to figure out how I can autocopy the configs using cmake with a 'make install', anyone knows?
User avatar
1v0ry_k1ng
Posts: 4656
Joined: 10 Mar 2006, 10:24

Re: A new artificial intelligent opponent - E323AI

Post by 1v0ry_k1ng »

why does this cause spring to freeze up so long in the first seconds?

are you going to reduce the amount of memory use?
User avatar
Error323
AI Developer
Posts: 237
Joined: 28 Nov 2006, 16:46

Re: A new artificial intelligent opponent - E323AI

Post by Error323 »

1) because its generating GOOD paths such that performs well ingame and I haven't come up with a faster algorithm. If you can spot improvements @ http://github.com/Error323/E323AI/blob/ ... finder.cpp line 88..157 let me know :D. Working on a caching function such that this is performed just once per map.

2) I'll release a reduced memory version soon... along with various improvements / bugfixes.
Post Reply

Return to “AI”