baserace
 

I created a proof of concept for what was discussed in the other thread. It seems to work as expected; I can kill the deliver actions when the constructs are finalized. All from one action test in the aiscript.

I've attached the test navs. There is zero bot support, so don't add any.

The test is done on the axis tactical ops and forward gate. If you manually build those and do a /g_action_info on the deliver action, you will see that the action type is changed from 7 to -1 (the effect of kill_action).

/g_action_info 2 for axis tactical ops
/g_action_info 4 for axis forward gate

By creating a fake construct, I can use the construct command in the map script to call the action test in the aiscript at any time.

EDIT: Moving Example to Wiki

I'll work on completing this map, it shouldn't be too hard. Although they won't be able to use any 'strategy' for building the bases, it should still be fun.

Re: baserace
 

**tips hat
**clap clap

you (and the rest of you waypointers) are fkn amazing...

one day i WILL get my head round all of this.. (gotta stop playing and start pointing :) )

thanks for you inspiration.

Re: baserace
 

Nice work, I just got one question. Is it possible to make the helper a different classname other than trigger_objective_info? Venice has too many toi entities and it just crashes if I add more in the map script.

Re: baserace
 

WOW - amazing work CrapShoot! Will be looking forward to seeing this in action. :)

Cheers!

Re: baserace
 

Thanks. It should be ready in about a week hopefully.

@420Blunt

func_constructibles need a TOI. I sent an email with a work-around for Venice.

Re: baserace
 

http://www.geocities.com/cameronbr/U...ementaward.jpg
for thinking outside the bots...
:clap::ninja::clap::blushing::clap::):clap:

Re: baserace
 

I have ZERO idea on what all this means but I look forward to it's completion and will install asap :)

Re: baserace
 

Attached is the first beta version for this map. Rename so it has the extension .pk3 and place in the fritzbot folder.

EDIT:

version 2 beta can be found here.

Re: baserace
 

High thanks for even attempting this..

i have a couple of demo's i did the other day, 1 the bot got stuck in the tunnels, where an exit was it got stuck on the left side of it and couldn't navigate outta the tunnel exit that happened twice. the other was really strange and have seen it a couple of time also, one allies in the trench one axis above the trench BOTH shooting each other BUT not even hitting each other, they were firing and hitting the edge of the trench and both ran outta ammo..

also is it me or do they possess a rather high acc. ? i have bot_aimskill @ 1 just for testing this, i even set it to 4 but there was NO difference...

one other thing i noticed (no demo) the cov.ops of ANY side sometimes stood infront of the crate to be collected and just stood there....

i have lowered the bot sight in the script to 1700 (hope you didn't mind) as 2000 they were shooting me before i even saw them even if i set the FOV to 110 they always saw me first...

hope that helps for now..

oh yeah not sure of you max bots you were testing but i looked into the baserace pk3 and saw you had assigned 16? bots so thats what i've been testing with.. maybe thats the problem and you expected a test of 6v6? i dunno..

great job on this though :)

Re: baserace
 

Thanks for the feedback.

I'm tweaking the pathing a bit still. I'm pretty sure I have that tunnel exit problem fixed.

My radius' for the steal actions were a bit big and I believe that was causing them to sometimes freeze in front of them. That should be fixed as well.

As for the aiming, I think they have problems adjusting when a target is at a different elevation :( It would need to be something Mal adjusts in the code.

16 seemed like a good number for the map. Feel free to edit the bot file if you want more or less.

I'll post a patch later tonight.

Re: baserace
 

version 3 beta / release candidate is here.

Changes:
improved pathing.
improved routing.
minor change to defense.
reduced sightDistance to 1700.

Known Issues:
bots will try to steal or recover parts that have been dropped even if they are carrying parts (AI related issue)
cvar out of range error if one team gets too far ahead (map or ET related bug I believe). Should never see it during normal gameplay.

Unless there are major issues with the waypointing, this will be the version that goes into Bindle's coming waypoint pack.

Re: baserace
 

FYI in version-2 tests the 3rd of 4-5 pass throughs
ERROR: G_Scripting: alertentity cannot find targetname "blue_forwardwall2"
I searched the script file but it looked OK, no obvious typo etc. Tried to reproduce but it worked fine...
The error occurred in single obj. when allies were up 2, allied fieldop stepped on the unbuilt forward fence pad (another bot was also carrying materials ahead of this bot so this might have been a "simultaneous" deliver but I doubt it). Allies had delivered maybe 5 already by that point in the map. I have saved the logs so if there is anything you want me to search for just tell me. I'll try out version 3 tommorrow.

Re: baserace
 

ahh, i know why that could happen. that's my bug :/

you're right, it's a near simultaneous deliver. I just need to move the alertentity stuff to the bottom of all those script blocks, rather than have them at the top...

Thanks for pointing that out.

EDIT: version 4 is here.

Changes:
Fixed near simultaneous deliver bug
Added a pathaction to one of the allied deliver routes

Note: the wait 1000 between alertentity and construct helper_ent in my original example is not needed (and probably shouldn't be put in).

Re: baserace
 

WOW quick update, ok i've got the wp file i'll get testing either tonight, or tomorrow morning whichever i'm looking forward to testing this..

update on sunday for you hopefully...

Re: baserace
 

Ver 4 tests:
- connection 22 - 157 goes through solid wall of axis mg/rocket tower at stage 3 construction
- IMO the long and safe underground path is too safe. there should be a camp above or near the underground dynamitable wall to kill the humans.
- CvOps seem to have the greatest problem in this map, pausing while carrying, going zombie in uniform near a pickup (and the cvop power-up hut deliver is ignored in favor of the medic's and the soldier's hut, so it may not be built until the others are maxed out)
- There was a setstate no entity warning occuring late in an long long extended-time map test, might be nothing important haven't reproduced it yet.
Overall quite good but I would only recommend it for single objective (or stopwatch which will run full maptime in any case).
- - -
"bots will try to steal or recover parts that have been dropped even if they are carrying parts"
yeah that one can be right funny :DLOL. Almost captured one screenshot of this one: material with a stationary bot in the center already carrying (all you could see was the green ! circle) and dancing around without firing 5 axis bots trying to recover the loose materials box as well as one allied medic. They must have been like that not fighting for 5 or more seconds before the bot at the center moved and the materails were picked up and the fight resumed.

Re: baserace
 

Well remember that they weren't even supposed to be able to play this map :p

Mal and I looked at one of the beta versions of the map prior to the release of the first version of Fritz ET and decided that a fair amount of time would be needed for him to code special features for it. So I think for them to be able to play the map as well as they are is a tribute to how much the bot has progressed.

And who knows, maybe we can all fly to ID headquarters and picket until Mal comes out and hands us a patch :D

I've been hesitant to add a real defense (it consists of 5 roams and 2 mobile mg42 camps atm) because I believe that it will negatively effect other things (related to the bots focus on goals). The only place I don't have a defensive roam is where you mentioned, so I'll add that and see how it goes.

I do plan on adding the mg42 camps this weekend as well as making them a bit smarter about which constructs to build first. There are some limitations for that, but I believe I can do some things for it. I just don't know what would be best because I've never really played the map...So if there are any requests for a build order, let me know.

Re: baserace
 

on a quick note before i test tonight, the mg42 in the tower can be skipped as far as it goes, not even human players often use it, the newly placed mg42 on the front of the central building is a new addition and would most probably be used, the tunnel definatly needs sokme def as i found once the axis tunnel on the right had been blown i could just use the tunnels easily to get crates and escape through the tunnels, rather long way roundbut VERY safe.. the onlt other thought i had at this time was about the engi defending the base, i.e. mines at the shaft which is neer the soldier hut, with the slope upto the shaft? it's common for human rngi's to mine there, also the other crate the slope behind that is also quite a common minefield, and also laying mines near and around the crates themselves..

will post more after i have tested this..

btw i am very impressed with your work on this, if i ever see you i'll get ya loads beers!!!

cheers for this.

Andy

Re: baserace
 

:detective:With 30 minutes the 16 bots don't complete the objectives anyway so you don't have to support the delivers for all 4 stages.
(1) the rocket guns are quite effective at only a couple of delivers each but that slows down the game steal-delivers after that.
(2) The forward fence tends to push the spawning bots to go single file out the center so i say de-emphasize that deliver, Also the bots defend best when they can fire back into their own areas, so I would avoid having them building the solid forward fence (kill the deliver at the barbed wire state unless you want to support climbing the chain-link fence in which case kill the deliver at the first deliver).
(3) An alternate to defending the underground wall would be to route through the center underground. I only saw 2 bots in the center during all my testing and I did a lot of running through the center. this could help the play in the second half when the bot xp slows down the success rate for steal-delivers. It would also allow for the bots to switch over to the other steal. (BTW did you ever test whether the bots can crawl out from under the medic hut?)
(4) Making the deliver paths less exposed would like help alot. i get the impression that it is the fight reaction that pauses the bots during the delivers and it takes them some seconds to return to the deliver action after the fight.
(5) Mines would be good too (note there is a bug of sorts when the engr building stage adds the pre-placed mines. Your HUD gives you 999 mines -minus what you placed but won't let more be placed saying that maximum was reached. Given the limits on Fritzbot stability re:>10 mines I would limit the mine actions to 3 or 4 JIC or don't support the engr deliver that far)
(BTW FYI testing long games got a G_spawn no entities error at 198 minutes of continuous play, the log seems to have a lot of airstrikes entities? maybe random RAM stuff)

And if somebody has time to waste, something else that would be an interesting research topic would be can the bots make use of the cvop hut teleport (i.e. an altroam on the pad and a secondary route at the underground cavern and not zombie or backtrack, or should the secondary route be at the teleport but its pathaction be within site of the cavern)

Re: baserace
 

version 5 is here

Changes:
Added routes for center tunnel
Added pathing under each base
Added defensive camp actions
Fixed an Allied Route
Fixed the Axis Assault pathing issue
Improved pathing

I think this will be the last of the updates for now (unless it's completely broken).

Thanks MetalHead and TomTom for the feedback.

Re: baserace
 

Thank you, CrapShoot.

Re: baserace
 

ok i'm gonna do more testing as i downloaded the v5 update a little late. lol didn't see the email notification :)

one thing i forgot to mention before was when the bots try to jump the gaps, some are easy but others require a speed boost to jump them, if they dont make it some get stuck 'trying' to jump back up lol, if there's 2 or more then they all jump on each other to get out, rather funny that :)

i have noticed the delay you put in for putting a crate on a capture pad, the only thing with that is that more often than not there's not another crate thats put onto the pad so there's no need to delay, the bot just stood there waiting then after a while finally put the crate on the pad, if a nenemy was around then it would be easy to kill the bot before he puts it on the pad, there still seems to be a prob with cov.ops mainly standing at the crate capture just standing.. i have only ever seen 1 other bot do it and it was a med. cov .ops don't snipe, or is that part of the bot ai you have to tell it where to snipe from? abv. engis don't lay mines yet, f.ops certainly like to spam arty and it doesn't appear to be a random(???) thing, they do like to drop arty at the enemy spawn which is normal for any human player to do.

p.s. quote tomtom 'With 30 minutes the 16 bots don't complete the objectives anyway so you don't have to support the delivers for all 4 stages.'

yeah with the current bot aimskill even on 1 it's hard to complete, i do set the timer to 300 just for this testing, but i think on a normal pub server it might prove interesting...

thats it for now, i'm gonna setup my server with the new v5 and test somemore, i'll post as soon as i can, only a few days 'till christmas so i'll wish you all a merry christmas and a happy new year :)

DONT try and rush to finish this if it has ti wait 'till new year then so be it :) i'll be very busy over christmas with the family..

Re: baserace
 

Nice work. I'm really surprised they work as well as they do, considering their limitations with understanding more then 2 steal objs at a time. :)

Re: baserace
 

Thanks Mal :)

@Metalhead

The map and bot teams are perfectly balanced. Even with all human players, I don't see how it could be fully completed if both teams are equally skilled (and have dedicated defense). You could set up the bot file to have one team more skilled than the other if you wanted to see more of the bases being built.

The bots will snipe without snipe actions, but I didn't add any snipers to the bot configs. Originally I had them in, but the snipers were dominating the map. You could change the bot config to include them if you want.

If the bots miss a jump over the gaps, they will spend about 1 second trying to get to the node across it before selecting a new path. I'll double check the jumps though and have any updates included in the coming waypoint pack.

Re: baserace
 

@crapshoot

yes my friend, i was looking too much into this, i understand what your saying having spent time with a clear head, i see both teams being as equal and both capable capture/delivery, i was looking at it a bit too much tactical, unlike goldrush for example you have attackers and defenders /allies/axis, BUT on this map there is no such thing as 1 side attack and the other defends.. thats where the human element comes into play, having our own individual idea of attack/defense/teamwork.

my apologies for looking too much into this..

i do appreciate you taking this map onboard and what you HAVE done is something once thought impossible but you've proven otherwise, yes i did add snipers and i saw your point you couldn't cross the field without dying quickly, so in conclusion on my part, i think this is a first class job, i watched the bots and played with them as closely as possible, yes when thry can't make the jump they DO infact look for that alternate way back, or jump up on a teammate to get out (thats still funny sorry :) ). so it that case the ONLY thing that might/does!?!? need tweaking is the placing of the crate on the capture pad, and that could be awkward considering what time allowance would be best, not sure if it's a bug in the map and the bots getting confused but the Cov.Ops (once a MED) stands at the crate just looking at it. while on about the cov.ops do they still check for landmines?

please don't take it as i'm being too fussy or grumpy about this because i am not! i've always liked baserace but to play this with bots i could only use *coughomni*coughbot*cough and i REALLY dont like them :cursing: .. so again thank you very much for attempting this i DO appreciate it :)

btw merry christmas to you all :santa:

Thanks

Andy

Re: baserace
 

Any and all feedback is welcomed. You've helped to make the waypoints better :)

Re: baserace
 

Lads, what version of baserace does this work on? The one I tried, all the bots stood still with no goals...

Adder

Re: baserace
 

It works with the final version. it sounds to me like the waypoint files may be in the wrong place. just rename the zip file to have the extension .pk3 and place it in the Fritzbot folder.

Re: baserace
 

ok 'http://homepage.ntlworld.com/andrew.carter847/demos.zip' download these demo's watch and see.. just incase rmember to put them in the /fritzbot/demos folder yes you knew that but was just checking lol..

i've noticed the defense position the soldiers use, also if the soldiers aren't there the engi's will take there place, if sol/eng comes by and the spot is taken it just plods around until the spot is free. :)

this is V5 i've tested but i could only test for a short while, hope them demos help?

Re: baserace
 

ugh, i think i know why that's happening at the deliver points :/

I may or may not be able to fix it. To make a short story long...the map script turns off the capture pad for 5.1 seconds to give the construction time to build. If another bot is going to deliver at that point and is close enough when the pad is disabled, it will get confused. It may take a couple days to get it sorted as i may have to toggle the deliver action off and on when the pad is disabled. That is going to require adding 40 entities and a bunch of scripting....but I'll give it a shot :)

Re: baserace
 

version 6 is here

Changes:
Fixed bots sticking sometimes at crates inside enemy bases (steal actions)
Fixed pathing at allied ladder to underbase

I'm still working on a solution for the deliver actions, but it's looking like I won't be able to (keyword limit for aiscript). If I can figure a way around it, it will be in the next waypoint pack.

Re: baserace
 

WOW didn't expect this turnaround, cool will test this tonight for you, looking forward to it, i'm glad the demo's helped...

Re: baserace
 

ok i haven't had chance to look at this sorry busy with the family :)

the only time i can get a look at this will be the evening of january 2nd or 3rd.. sorry for delay but will post as soon as i have tested it for you :)..

seasons greeting to you all..

Andy

Re: baserace
 

I had just installed Baserace on my server and it sets 8th in a 10 map rotation. For some reason the map doesn't go on to the 9th map. It stops the rotation. ET then restarts the server rotation with no bots on it. Unfortunately I had to remove this map and now everything works fine.

I've never actually seen what happens on this map that causes it to reset. But I do know that something in this map is causing it.

Has anyone else had a problem with this??

Re: baserace
 

I'll take a look. Do you know if it happens if the map goes into overtime only; or all the time?

The initial guess is that campaign mode wants a winner at the round end, but he sets that value to -1 if the score is tied....

Re: baserace
 

Re: baserace
 

there may be a way to fix it, not sure yet. the more I think about it though, I can't imagine that the game doesn't account for overtime; more likely it looks for round time and a valid winner value before ending the round....unless this is a problem on all dual objective maps in campaign mode...

I'm just setting up a server now to test some things.

Re: baserace
 

no probs with baserace here as far as map cycling goes

Re: baserace
 

well the exact crash is:

G_Spawn() no free entities

this is bad. it never showed up with 9 v 9, but I guess at some point it would have. There is a maximum of 1024 entities allowed in the game at a single time. During normal gameplay, new entities are spawned (like grenades, airstrikes, gold crates, etc). Given that the default map has roughly 880 entities at initial spawn, I can see where the limit would be reached very easily (google search shows it has). Given that I needed to add 46 to provide the basic support for it, I think that this map is dead :(

I've yet to see it happen 9 v 9 and lower, so I guess use at your own risk...

Re: baserace
 

Saw it in time extended testing that i mentionned back in a post on page 2. The log file showed a lot of air strike entities in the entity list dump. So I wonder if restricting the airstrikes or something in some way might keep the map alive?
BTW does anybody know how ET does garbage collection in the entities?

Re: baserace
 

no real garbage collection that i know of, they just use the entity and free it after it has served its purpose. it's just that a regular airstrike can spawn a bunch of entities (one for each pending explosion).

i guess i could go looking for entities that can be removed, but really I have no clue what the magic number is. i'll play around with some things this weekend, but I don't think we could ever call it stable enough to put on a dedicated server....