Hope this question to developers is not asked in vain

_X_Avenger · 2003-02-20 06:45:47 UTC

Just wanted to know if you were still planning on including bots. – Almost everyone will desperately want ET for the MP online play, but just wondering about bots because 1) we have not heard if they were still a go, and 2) they could still be extremely useful to many RTCW fans.

Locki · 2003-02-21 00:40:47 UTC

A great deal of effort has been invested in the multiplayer BOTs for Enemy Territory, but whether they make it in to the final game or not will ultimately depend on whether they work really well in time for release.

I feel very possitive about them, but like every other Enemy Territory feature, we’ll only include something in the final release if it meets id Software’s high standards.

Paul/Locki.

_X_Avenger · 2003-02-21 07:48:24 UTC

Cool –

Thanks for laying that question to rest; we now at least know that BOTs with ET are still a possibility. It is nice to see a developer concerned with taking the time needed to release nothing less than the finest product. Regardless of BOTs making the release, ET’s gonna rock!

Developing BOT ai sophisticated enough to handle a game of ET’s complexity must be QUITE an interesting challenge, especially considering the difficulty human intelligence sometimes has with teamwork – especially on the pub servers.

funkyjunkie · 2003-02-21 08:29:06 UTC

i really hope bots ll be included

just one question, why does it depend on id if bots ll be released or not. thought activison says what to do!

@locki is ur answer just blabla to clam down enemy territory fans, or is there really a chance that bots ll be part of et?! what do u think about it?

funky

O==[]:::::::> webmaster of: <:::::::[]==O

------------ RTCW online.de -------------

chavo_one · 2003-02-21 19:11:11 UTC

If you guys can get bots working in the MP mode, then something sounds fishy as to why the SP mode was canned. The reason I say that is if MP has bots, then it is simply a SP game that multiple people can play! Wasn’t the bot technology the biggest and most challenging addition to ET over RtCW?!

DJ_Art · 2003-02-24 21:17:39 UTC

Heh, everyone’s so obsessed with these bots!

Thinking as a gamer I would prefer to see no bots rather than lower than “high quality” bots in a game.

You have to be realistic on some matters. Real-kicking ass AI is a tough matter. Something less than perfect would end up being boring pretty fast.

Oh no, big question here! and you’ll ask yourself: “What the hell is a perfect AI ??” The last good answer I saw to this question was a small blonde AI falling off the 10th floor right into the water :> :eek3:

oh well, be cool! bots or no bots, the servers have to be full of humans, not full of a bunch of perfect engineers that always follow the orders by book :>

chavo_one · 2003-02-25 04:13:43 UTC

You have to be realistic on some matters. Real-kicking ass AI is a tough matter. Something less than perfect would end up being boring pretty fast.

I have seen ass kicking AI before so that in itself is possible. The real challenge in an ET bot AI would be the teamwork aspect. Each bot has to have a role in a larger operation. His single goal is not just to kill and survive.

And I’m not here pushing a server full of “a bunch of perfect engineers that always follow the orders by book”. I would love the ability for smaller clans to practice strategies against a bot squad. And I don’t think that requires the “perfect” AI.

Also, my original point was, why cancel the SP mode if you are able to accomplish the most challenging game code in the MP mode?! (and then give it away for free) ::smells something fishy:: :???:

DJ_Art · 2003-02-25 08:02:03 UTC

This is why I talked about obsession. Why is it necessary to link facts here and there ?? What do MP bots have to do with SP release ?

It’s obvious, you mean that all today’s single player games are based on how good the AI is. True. But here, we’re talking about 2 different games. SP is so much different than MP. SP bots have to behave so much different than in MP.

There is big difference between a true single player game and this: “if MP has bots, then it is simply a SP game that multiple people can play!”.

_X_Avenger · 2003-02-25 08:26:37 UTC

This hideously long post is not meant as a criticism of anyone above, just my opinion–

Look at UT’s BOTs. They’re designed around a relatively basic recursive algorithm – they were able to play Assault and CTF modes just fine – they could cover each other in small ‘fire teams,’ give each other backup and defend the flag carrier from pursuers; generally they played pretty well. When you’d open up the files containing the algorithm, you could see how UT’s BOTs work – same logic that goes through a human player’s mind: defend or attack; if defend choose position, or roam; if attack, go right or left or under or over; towards closest players, or group of players, or go other way; use best weapon depending on opponent type and #, or avoid enemies and sneak by putting objects between self and player, or be outright aggressive, or sneak and be aggressive if close to the obj and only a few opponents are there; wait/call for backup when teammates are near; attempt to sneak when all dead or wait for respawn, etc. UT’s AI was not too complex, yet worked effectively.

ET’s BOTs only really need to give ammo and medpacks, revive closest wounded players to each medic who aren’t too far away or way back near spawn, and plant dyno – they would also need to break into fire teams each pursuing a mission objective, but those teams and their objectives could be set up in advance. They would also need a process for deciding whether to wait or charge, and how to react to the opponent’s basic habits, e.g. reacting to camping, roaming, charging, or other possible enemy behaviour. Sub-algorithms could be designed to handle each type of decision – I remember working on Turing Machines while studying grad Logic – seems that these simple decision-making devices used en masse could at least get the BOT AI part of the way to where they need to be, but I’m far from possessing any in-depth knowledge of the challenges involved in programming BOT AI.

Programmers need to know their maps and know how a player’s mind works when he/she plays – as players themselves, Splash Damage’s programmers have this knowledge, which translates into programming the BOT’s decisions. As with human players, BOTs would need to know the objectives, and the logical order for completion, plus feasible combinations that order might take – all of these things could be done with a recursive algorithm similar to UT’s, just significantly more complex and accounting for all the possible decisions the BOT should/must consider at each point along it’s route. And BOTs can be designed to move as a group; choosing a route and changing as needed. Not a simple task, but still extremely possible. It could also be possible to allow the BOT to modally choose between actions that carry varying levels of risk so that the BOT is not so predictable. For example, BOTs could be programmed so that on ICE, one might break away from the main play and camp the radio room, or on Beach, BOTs might boost each other over the wall a specified percentage of the time. Another possibility is that BOTs could choose from a variety of locations to camp the objective, so as not to be in the same spots time after time. And a wide variety of decisions regarding the groups/routes/and level of caution the BOTs choose to attack with could also be possible, e.g. some laying down covering fire and sniping while others advance, or waiting while the PF goes forward first to clear a room BOTs ‘remember’ as a point they were ambushed on the previous attempt.

Most, if not all the decisions a human player makes could be incorporated into the BOT AI IMHO, provided the algorithm is complex enough to take all or as many possible options into account depending on the circumstances at the time. Playing the game is really just a collection of decisions each player makes along the way according to what is best for him/herself, balanced (hopefully) with what is best for the team, under the circumstances.

ET’s AI is an admittedly complex task, but not impossible for the willing and able developer. Here, Splash Damage seems not only able, but also very willing to at least make an attempt to design a BOT AI that can handle RTCW’s basic tasks to at least some degree. They don’t have to make AI perfect – after all human intelligence isn’t; if it were, game play would be boring, nauseatingly predictable, and uninteresting – 24/7 Beach servers survive because no two games are alike; no set of players’ judgments are always perfect. For example: sometimes the PF takes out 4 teammates killing one guy – far from perfect. Only the clearly wrong choices a BOT commonly repeats need to be eliminated; the PF can’t always take out five guys to kill one enemy except under certain circumstances. BOT AI is most perfect when imperfect. Splash Damage’s team may pull this one off; at least give them credit for believing BOTs are still possible and not giving up.

Crapola, I’ve just written a book. Forgive me.

_X_Avenger · 2003-02-25 08:56:10 UTC

No – I bet Locki just set aside some time to type out a B.S. answer to placate a group to which he is not obligated to answer. He’s probably not busy on anything else and has nothing better to do.

ROFLMAO

Sorry to pick on you, man, but you left yourself open – Locki could certainly not be expected to post a reply saying “Oh yeah, I lied – BOTs were never a planned feature – we just wanted to see if we could manipulate consumer confidence through a fraudulent and absolutely false statement. Have a nice day.”

Jeez if he says they are still planned, they’re still planned; they are not going to flat out lie when they have the clear option of not responding at all. Have some trust.

fretn · 2003-02-25 15:32:38 UTC

CoBraLorD and I are working on bots for rtcw

here is a small demo:

http://users.pandora.be/sodoff/files/bots/surprise.dm_60

atm they can:

do some basic shooting, and they can follow a route to find certain points in the map.

expect some more in the future

fretn

chavo_one · 2003-02-25 15:49:07 UTC

Since I’ve been labeled as “obsessed” I have to reply once more. Avenger made a very clear case as to what is probably involved in designing the bot AI for ET. It is surely a monumental task since every map needs its own strategy. (and what about all the custom maps to be made by the community?)

My point is still valid. Activision is a public company that has to make money for its shareholders. They are out to make money plain and simple. How can you expect them to release ET for free if it is anything more than an advanced mod for RtCW? You see expansion packs for games all the time that cost less than the original release. Why isn’t activision releasing ET in this manner to at least regain some losses? I’ll tell you this, it’s not because they love you and want to give you a present.

JungleDave · 2003-02-26 00:33:12 UTC

If they cancel out the bots for ET. I would hope they will release the code in an SDK for the users to finish. I play DOD & Counter Strike and the user created bots are down right nasty to play against. They hunt you, plant bombs, defuse them, capture flags,etc. Someone creates a custom map and bots can be added using waypointing and a command system. The person running the map decides how many players. The humans join in and the bots fill up the rest. RTCW would be smart to follow this trend in game design.

_X_Avenger · 2003-02-26 16:52:43 UTC

Hey Fretn

That bot looks pretty nifty keep up your work!

BTW I like your modesty – ‘basic shooting’ – LOL – they look like headshot experts.

A_KillerMan_F · 2003-02-27 16:56:05 UTC

Hey guys, I am an avid player…actually, that’s a joke. RTCW has become my freakin life. The creators did an excellent job in making such a creation, near perfection!

I direly await the arrival of ET. And I was searching on google.com for anything on rtcw, and found this discussion. Personally, I was wondering myself about the presense of bots.

Keeping this simple, I know Avenger, not personally, nor does he know me, but I’ve seen his/your clan several times. BLack and blue I believe. We were clan Xtreme, X[KillerMan] was my name and tag. We changed. Basically, I registered for this forum just to get involved in the topics as well! :bump:

When you create the bots, will you worry about the ping issues? I live in Idaho, (sadly) and that is all the way across the nation for nearly all servers. I ping over 150 in all cases, except Utah and California servers (as you would imagine). If there are bots runnin around in servers with 0 ping, regardless of how “perfect” you guys make them, they will tend to be better players anyway. If you could make it perfect 50 ping, that seems do-able. I have played this game (sp too) since a month after it came out. I even played when I had my 56k for 4 god damn months. It was and still is that addicting. I have been in a veteran clan TeamMerlin, back when they had respect. And my new clan is earning its place. We fully intend on getting the Expansion and owning it as well.

The thought of making the bots is great. I figured besides Avenger, you might want more comment from a person who plays this game well over 40 hours a week:) LIke i said, obsessed, you created the best game ever! Sadly so, playing so much, I dont get money:( I wish i could get paid to play, but I must have missed the signup. Originally, my fellow teammate (who was leader of Teammerlin when i was in it with him) were and maybe are still planning on hosting a huge tournament towards end of summer (maybe August or July). I had actually received “go-aheads” from local sponsors who agreed to help fund it and make servers and help donate for the grand prize, 100K. My plans for hosting this tournament were because www.cyberathlete.com has played their tournaments for quake and all those other ghetto games (cough* CS *cough). I tried joining their forums, but they wouldn’t let me for some odd reason. If they wont host a tournament for this game, I will. My buddy who was helping me with this task has the recourses, money, and sponsors himself (including IBM, not big anymore, but we have/had big names). Honestly, i wanted to ask for a “fronted” copy of ET and see if we could base the tourny off of that or off of original rtcw.

O wow. I got way off topic, I guess i was introducing facts not in evidence:( Back to the bots. Make them perfect if you like, I’ve seen and know players who are bombshit and have been called for cheating before, your bots wouldnt make a difference. But the Bot config for the UT ones, that sounds logical, give them direction, names, class, and what to do. Or maybe even for practices. Have a clan play against them, just like the headshots only mod for shrub servers, let us get better against THEM.

I talk a lot, im really sorry, but I always play this game, I commend those who made it on an extremely well done job! :drink:
Thanks
A]KillerMan[F

Nothus · 2003-03-05 20:56:35 UTC

Hey!

I’d just like to ask really quickly:

How did you get the bots working??

Did you just use the q3 bot src as a base??

Are you writing your own code from scratch??

Are you using the CastAI??

Just wondering.

btw: haven’t seen your demo yet, so forgive me if I sound naive. I’m at school ATM, but will watch it later tonight for sure!

Yesterday, as an experiment, I modified some of the default q3 bot funcs, and was able to get a bot to connect properly to the server as a client, but without the special AI it needs to select teams and class, it would just sit as a spec the whole match. A start yes, but I’d like to know if there is an easier/better way.

Either way, I hope you continue your work!

I hope to perhaps work on a bot for RTCW myself, if one isn’t provided by ET…

DJ_Art · 2003-03-05 23:44:27 UTC

There are a lot of guides and tutorials and example code for Q* bots out there on the Internet. Make sure you begin with a “perfect” search all around the community sites and of course with google.

Q* means for Quake2 and Quake3. Q2 may sound a little bit strange but IMHO the more example code you have the better idea on how to code bots you’ll get.

(And I think that these are tutorials and example code from real professional programmers, working in these familiar game development teams).

A_KillerMan_F · 2003-03-07 04:50:00 UTC

I just watched that demo of the bot. 2 words, OMG! Lol. I think it took headshots for a major in college. But, if you fix all them bugs, i think you will have yourselves another masterpiece, just like the original game itself! and then you can party :chef: , and drink beer with a friend :drink: and celebrate all the awards coming in like rain!

Now, I might as well use this time to ask a question. I wish to be a gaming designer such as you guys, maybe work for id software or other places. Anyway, with that in mind for my future (cuz games are already my life, specifically this one! lol). What fields of study should i take in college? And have any of you gone to ITT Tech to get your degrees? I want to work with audio technology, but autocad looks to be coming into the gaming world. And character design seems interesting as well. Based on preference, pay, and college courses (dont care how tough they are, i’ll still get through it well), what would any of you guys recommend?

thanks

fretn · 2003-03-09 08:39:24 UTC

we used the fusionbot (bots for the qfusion engine) code (that’s is based on the acebot code for q2) as a base. But atm there’s nothing left of the fusionbot - code

fretn

Nothus · 2003-03-11 18:21:07 UTC

Interesting.

FYI:

Just hit a brick wall on my idea of using the q3 bot source thats already in Wolf: it WILL connect a bot, but it crashes the .exe at seemingly random times, on an empty, dedicated server.

Also a lot of AI related trap calls/funcs seem to be broken or disabled, I had to comment out, or replace a load of them just to get the bot to connect period.

Its hard to tell for sure, since everytime it crashes, it doesn’t do so in the game DLL where I can debug it, but in the .exe, with no msgs or warnings.

Jeez.

I seriously hope that Splash Damage will release those bots in ET - or at the very least (even if they do suck), make it possible for the community to take them up and finish them. Leave them in the code, tucked away from the average gamer if you wish, but in plain site for coders.

Please.

I for one will gladly take them up.

btw: I’ve heard before that its impossible to add bots to RTCW using the q3 bots structure - does anyone know what all they did to make this so? And is it something that can be reversed, without access to the .exe?