|
Re: POD-bot back into shape.
Just curious how the collect experience deal works with the bots. Just wondering how it runs and how it collects experience and to what extent lol i've played with them and a number of times they just sit in corners for like 5 minutes jumping around and stuff not moving around. Good bots though. Thanks for making them available.
|
Re: POD-bot back into shape.
These bots simply remember the waypoints where they were last killed and the position of the killer. They weighten all their waypoints with an integer value describing its danger, and use this information to look at the appropriate positions when wandering around.
Your navigation problems certainly come from bad waypoint files. Ensure you're using the right waypoints for your map. Note that in CS 1.6 some maps have been changed (de_aztec, de_piranesi, etc.) and some waypoint connections have been broken. Look in the waypoints folder, there are two versions of the same waypoint files when the maps differ in CS 1.5 and 1.6. Just rename the one you want to use with the appropriate name, and your bots should run ok. |
Re: POD-bot back into shape.
Alright thanks.
|
Re: POD-bot back into shape.
1 Attachment(s)
The bots lag after a map change and buy extra weapons for no reason.
i have demos of them doing those things. my system specs are: Processor: 400Mhz PII RAM: 192MB SDRAM PC100 HardDrive: 20GB Video Card: Nvidia Riva TNT2 Model 64/Model 64 Pro 32MB Monitor: Pavilion M50 Sound Card: some incredibly old Intel sound card VidCard Temperature during test: 38.0 Celcius OS: Windows 98 well this is all the info i have but nothing about radio problem yet. they just don't use radio during round, only beginning of the round. |
Re: POD-bot back into shape.
Mmmm this may be a stupid question to ask. Are the waypoints in use set with the exact name of the map? If it is so then the 1.5 ones shouldn't be in use. The 1.5 ones are set as de_aztec (CS 1.5).pwf and the such while the others are just de_aztec.pwf lol I know it's probably a stupid question and possibly has been answered within the 16 pages of posts but I do not have time to flick through each one I have a few assignments to do and I am trying to load these on computers at the school. *whistles innocently*
|
Re: POD-bot back into shape.
it will try to load the waypoint file with the same name as the map. so for de_dust.bsp it'll load de_dust.pwf
|
Re: POD-bot back into shape.
The bot lags because it builds a visibility table for each map it discovers the first time. It lasts less than 1 minute and the data is then saved in the bot's experience files. It won't lag like this the next time you run this map.
And if the bots buy extra weapons, perhaps that's because they have cash to spend ? :) I'm working on a few additional fixes anyway. |
Re: POD-bot back into shape.
Quote:
and about the radio usage, there was indeed a bug (in my own code) that would make the bots react to radio commands when issued by a player but not to radio commands issued by bots themselves anyway there will be yet another release. I've sorta fixed stuff in the camping behaviour too, which was incredibly broken in the POD-bot 2.6. changelog: Quote:
|
Re: POD-bot back into shape.
Gogogo!
|
Nice conversion....PM
Sup all.
First off, nice job on converting the bot to metamod PM 8) I have some interesting things that are happening during play that I thought I would note: 1.-Bots tend to have problems with navigating slanted doors. Example: de_dust. Not sure if this is due to bad waypoints. 2.-Bots sometimes walk and or run a small route in an endless loop. 3.-Bots are buying extra weapons they allready have. Somtimes buying an extra weapon and leaving it on the ground in front of them. Rarely, they will buy an extra weapon and actually pick it up. That is what I see so far. Nice work PM and everyone else who has contributed to this bot. Until next time...Cheers! |
Re: POD-bot back into shape.
1+2: Bad waypoints.
|
Re: POD-bot back into shape.
PMB,
Server crashes on Linux have fallen off dramatically. Once I figure out the errors I will let you know. Thanks for all the hard work. Cooley |
Re: POD-bot back into shape.
First of all: Superb job!
Play the bot as a listenserver and there have been some amazing things happening... :-) unfortunately the bot seems to put a lot of load on the cpu because on large maps like cs_crest the bots move in slowmotion! even if i reduce the mumber of bots to 6 (usually i use 10) they take it slow. My system: AMD 1333; 512 MB, KT133 Mainbord, Win 98. Any Idea how to speed em up? On small Maps the Bots go allright! |
Re: POD-bot back into shape.
Was it the first time you loaded the map ?
If so then the high lag at start is normal ; it's because the bots are building a visibility table among all the waypoints. Once this table is built it is saved back in the bot's .pxp files and is just reloaded the next time. The lag corresponds to an intense TraceLine activity during the building of the table. Once it's finished (about 1 minute or so) then the bots should run fine. |
Re: POD-bot back into shape.
PMB can you create another thread for when you update Podbot? Also can the [POD] tag be changed to use your own. Suckup part: Excellent job on the port. :D Side note: for some reason my Statsme works properly again. Also is that a smile on the penguin. Mikey I think he likes it.
|
Re: POD-bot back into shape.
@PMB
Hi! Been testing for a while: With 8 bots i got good results. on the big maps it took quite a while before the slowmotion effect ended (5 rounds or so). but then i had the impression that the bots speeded up every round! good! So I started again with 10 bots. That was a mistake. The game fell into a kind of super lag state (like 1 second movement, one second standing) and the bots appeared to be disoriented. i kicked two bots but without effekt. So I exited and restarted the game with 8 bots. same map everything ok. round ends, mapcycle loads the map again (8 bots), superlag state is back! Strange thing. Maybe i messed up the experience file by adding two bots.???:( Anyway second question: you said you wanted to publish the poxbot waypointeditor or its code? I m very much looking forward to this as waypointing with the old one is a pain. could you please post a dwl link to it? thx |
Re: POD-bot back into shape.
Does this podbot already include the fixes from podfix?
|
Re: POD-bot back into shape.
@Abu Amok: Austin's waypoint editor is already included with this bot, and if you downloaded it then you also have the source code since it is included in the .zip file. Look in the ClientCommand() function of dll.cpp for the waypoint editing menus and command code and in waypoint.cpp for the editor graphics.
About your lag issues, it is completely abnormal that with such a hardware as yours you experience such a lag with the bot. Try deleting all your .pxp files and double-check your waypoints. Bad waypoint files may cause the bot's A* pathfinder to fail and expand in all directions searching for a path that doesn't exist (and raising your CPU to the max). Be sure there are waypoints EVERYWHERE in the map. But try deleting your .pxp files before anything. @sPlOrYgOn: no, nothing from podfix has been included, but instead bugs have been fixed directly at their root in the bot code. PodFix was actually a "patch" that corrected wrong symptoms by applying their opposites on top of them. But in this release, the bots do plant and defuse the bomb correctly each time, they don't crash anymore with the hostages (well, not for me at least), and they snipe the way they were intended to (although the PodFix sniping algorithm is much better, I concede, but my point was not to improve POD-Bot's one but rather to fix it. So that it works.) |
Re: POD-bot back into shape.
Pierre,
I was running the latest build of your podbot-mm and it had crashed on the map fy_aztec. Here's the error: hlds_amd: bot.cpp:8095: void BotThink(bot_t *): Assertion `(pEdict->v.v_angle.x > -181.0) && (pEdict->v.v_angle.x < 181.0) && (pEdict->v.v_angle.y > -181.0) && (pEdict->v.v_angle.y < 181.0) && (pEdict->v.angles.x > -181.0) && (pEdict->v.angles.x < 181.0) && (pEdict->v.angles.y > -181.0) && (pEdict->v.angles.y < 181.0)' failed. Could these errors be due to bad waypoints? I hope this helps, Dethpod cstrike.rusin-central.com:27015 (Public) cstrike.rusin-central.com:27016 (Private) |
Re: POD-bot back into shape.
No, they aren't.... wtf I thought the assert() issue was only a problem with ideal_yaw ? and I thought it was fixed already ?
I will remove this assert() call. There's definitely something wrong with it. Maybe the expression to evaluate is too long. *edit* done. Test this new version instead: http://racc.bots-united.com/releases/podbot.zip |
Re: POD-bot back into shape.
Pierre,
Here's another error that happened: hlds_amd: bot.cpp:1869: class Vector GetBombPosition(): Assertion `vecBomb != g_vecZero' failed. This was on the dust2 map. People are really liking this version though. Great job. Dethpod cstrike.rusin-central.com:27015 (Public) cstrike.rusin-central.com:27016 (Private) |
Re: POD-bot back into shape.
Hrm. From now on I will compile the Linux binary with the -DNDEBUG flag. That will get rid of all the assert() calls. The bot should then run much better in Linux but it still won't tell me why these asserts fail, though.
Try this one (we'll finally end up with a satisfying result someday, won't we ???) http://racc.bots-united.com/releases/podbot.zip |
Re: POD-bot back into shape. [amx?]
I have a stupid question that may already have been asked before...
but first let me just say I **LOVE** the Metamod version of Podbot, it's awesome!! So far it's my favorite bot out there, the most "human acting" (by that I just mean that a n00b like me sometimes has a chance at killing one and it doesn't just headshot me right away like some other bots). Anyhow, my question - I've been running podbot_mm with amx, and I've only got maybe 1 or 2 amx plugins other than what's included with the standard load. My server crashes fairly frequently with this combination. I get some kind of sz_getspace and sizebuf_t errors ???:( [I have no idea what that means]. Thanks in advance for looking and more if you have an answer! -Mike |
Re: POD-bot back into shape.
Does the bot run well WITHOUT amx ?
What are your system specs and how many bots do you run at once ? Is it a dedicated server or a listen server (LAN game) ? |
Re: POD-bot back into shape.
@PMB
Hi! I still have my Probs with the slowmotion issue so either my comupter has a problem or the waypoints i use are crap. i'm going to redo them which leads to my questions: how does the A* navigation work? I have only just begun to read your Racc page so maybe the answer is there. If not, a short statement would be interesting... (actually the longer the better if you've got the time...:-) Are the terrorist/CT important wpt still needed? or does the AI first calculate the shortest path to the mapgoal point and later on the safest? So far I separated the Waypoints at the spawnplaces to divide the bots into groups that go different ways or to assign special tasks to single bots (if they spawn on a specific spawnpoint) maybe thats confusing the A* navigation and I should rather evenly distribute the waypoints on the map? Thanks for your help |
Re: POD-bot back into shape.
If you're interested in reading about A*, the best documentation I've ever found was this article on Gamasutra:
http://www.gamasutra.com/features/19990212/sm_01.htm It explains the differences between pathfinding algorithms in such a simple way even a kid would understand. The article requires a login to Gamasutra ; you can safely create an account, they won't sell your email to spammers like Gamespy does. And there are tons of interesting articles to read there. In the end I believe waypoints should be distributed evenly all around the map, in all places a player could go, with perhaps a little more density on the most tactical zones and less density in wide open spaces where nobody ever goes. It's the A* pathfinder which will evaluate the best path given the domain knowledge the bots have of the map and the danger of each area. It won't necessarily be the shortest path always, but rather a weightened compromise between shortness and safety. |
Re: POD-bot back into shape.
Quote:
I haven't tried the bots without AMX yet... I'm running a Pentium 4 2.4GHZ with 512MB ram on Windows XP Pro, I have a minimum of 5GB free on any given partition. Using that last version of CS 1.6 that Valve updated us to this past weekend. I generally run about 5-8 bots on a listen server. I haven't seen a performance slowdown, it usually just crashes. Thanks again for your prompt response! -Mike |
Re: POD-bot back into shape.
aww man..
address is required! I have security paranoia! Great job on the bot 8) |
Re: POD-bot back into shape.
This is my first bot experience using this bot. I loaded up podbot on my opteron server with 24 bots, and the server is crashing about every 10 - 15 minutes. I am using AMXX along with podbot. The debug message is:
CRASH: Wed Mar 17 21:00:20 CST 2004 Start Line: ./hlds_amd -game cstrike +192.168.0.13 -port 27015 +maxplayers 32 +map de_dust2 +exec banned.cfg -autoupdate -debug -pidfile hlds.1958.pid Using host libthread_db library "/lib/tls/libthread_db.so.1". Core was generated by `./hlds_amd -game cstrike +192.168.0.13 -port 27015 +maxplayers 32 +map de_dust2'. Program terminated with signal 11, Segmentation fault. #0 0xc8adaad1 in ?? () #0 0xc8adaad1 in ?? () End of crash report I would love to use this bot to stress test my computer, its just too bad it crashes so much :( hondaman www.hardgaming.com |
Re: POD-bot back into shape.
Remove AMX and try again.
Does it crash without ? If not, then not my fault. |
Re: POD-bot back into shape.
Yep. You're right, I don't know why I didn't try that the first time around. PODBot is rock solid and super-stable without AMX. Crap. Guess I'll have to try to narrow it down to one of the millions of AMX plugins that are present. Thanks again for the PODBot port! It's amazing!
|
Re: POD-bot back into shape.
Indeed, it does seem to be AMX. It hasnt crashed for 1 1/2 hours without it, using 26 bots.
Here is my thread on the AMXX forums FYI http://www.amxmod.org/forums/viewtopic.php?t=351 hondaman www.hardgaming.com |
Re: POD-bot back into shape.
I dunno whats wrong with you guys but I can run amx fine with podbot.
plugins i use are gore and warcraft3 along with all the default plugins. NEVER crashed. |
Re: POD-bot back into shape.
|
Re: POD-bot back into shape.
Okay...I stripped back some of the AMX plugins and podbot_mm did run without incident all night. 6-8 bots and about 4-6 real players at any given time. I also shut down a few metamod plugins...hm. this is going to be a pain to track!
I shut down the cs-stats metamod plugin as well as the funwhatever plugin that come with amx and left the others on and it ran fine...going to start adding them back in one at a time. Will report back if I find the one causing the mischief! |
Re: POD-bot back into shape.
I ran it with JUST the stats module, and it crashed for me. Im running it now, 26 bots, without it atm to see what happens.
hondaman www.hardgaming.com |
Re: POD-bot back into shape.
Quote:
one is running win98 192MB RAM TNT2 and the other runs winXP 512MB RAM GF4Ti4600. I'm running only podbot, amx, and statsme. amx has 2 plugins gore and warcraft3 v2.22.11 |
Re: POD-bot back into shape.
All my hotkeys get messed up when I use this. Like the GUI buy menu, I can't use rebuy previous, etc. And no.. I don't have my caps lock key on.
|
Re: POD-bot back into shape.
I think there is a conflict with statsme. I disabled stats me and the server was rock stable. Browsing the statsme forums suggests the conflict but I did not find the threads. I had forgotten to rank players by name and not authid which can cause problems. It looks like statsme and the stats plugin for amx are problematic for the podbot.
Cooley |
Re: POD-bot back into shape.
1 Attachment(s)
I will look that up, there's maybe an explanation.
*edit* would you mind trying this DLL ? (Windows only) If it works I'll build the Linux one and make another release. |
| All times are GMT +2. The time now is 11:53. |
Powered by vBulletin® Version 3.8.2
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.