I have this worrying feeling that the guys at Ubi think the current behaviour of the AI is "good enough" and doesnt need investment / development compared to other areas of the game.

If you play this game for the coop T-Hunt then hopefully you know what I mean. This is not about the spawn bugs but instead the AI as a whole. Today I found myself constantly thinking about what features I think the AI need and here are the distinctions I arrived at:


AI should primarly be attempting to achieve something i.e. holding hostages / stealing money / setting up explosive device etc. Everything they do should be directed to this end - as aposed to trying to kill the player team.

AI should primarily defend with a core team around the objective and fire teams patroling and scouting a perimeter. They dont give a sh_it and the player team aslong as they dont stop them achieving whatever the hell.

Once AI becomes aware of the player team presence it should become more alert and possible dynamically plan an ambush between the player team and the objective. Generally though it should only attack if it has a good chance and more frequently should fall back and camp.

So most of the time the player is attacking a well defended position using stealth / smoke / flash / grenade to give it that split second enough time to clear and progrss to the core objective.

AI should only start hunting and attacking if: they can first pin down the player team by covering all routes to the objective, and can then use smoke / flash / grenade to attack them once pinned down.

By attack I mean move towards the player team instead of laying in wait.

So effectively the player team has to constantly avoid getting pinned down, use stealth and evasion to get to the core objective and most of the time be attacking and entering rather than just camping and shooting AI as they aimlessly walk towards them.

Sorry I cant express this better but I really think its the core problem with the vegas AI and somehow I still hope that UBI will do something to make this change. I still believe rainbow six can be a truly great game and set a standerd for online coop if they stop shrugging off real AI coding and forget the gizmos.

I totally agree with your point of view!
But I personally think that it'll be too big of a job to get it done in this game... We will have to wait for the next one to hope this might be done correctly.

(I personally think that they ended this "Vegas" story because it was going nowhere and needed a fresh start... Well I hope...)

That would be what a game should be from a Dev that has pride on his work or a company that stands behind their product but, alas... we will never see that from the likes of UBI since all they are looking for is Milking that tom clancy cow (so to speak).

If there ever will be a new rainbow game I bet Ubi will drop TH altogether as it seems to be to difficult for them to code "correctly" as it is now it's to "dysfunctional" for them to care to have it in any other game developed by Ubi. Only guessing of course but I would not be to surprised if my guess turned out to be correct.

I also agree to what Want2Snipe says above, any dev that takes pride in hes/shes work would make sure it worked pretty much as you describe. Ubi seems to be rushing out everything they do, to bad for Ubi that they don't see that they might be rushing to there own grave, I'm not sure they are, but my common sense say they might be. However as mainstream gaming has become I'm not sure that "unfinished" games matters all that much when it comes to sales, people will buy about everything by judging by video reviews (people; stop taking GS and IGN seriously, they are financed by companies like UBI), time will tell if Ubi really are rushing for there grave..

Gaming has become like renting movies, you rarely check reviews and trailers before renting, you simply look at the cover, if it looks like a cool movie and theres some famous actor you like in it you rent it (at least I do) It's the same story for gaming, if the "cover" looks good you buy it...
* Note: "cover" being the whole marketing package.

I completely agree with everything said already, especially the OP's points.

I also think the AI should perform like realistic human beings and not super-humans. They often fire from extremely long ranges with pinpoint accuracy while running as fast as they can. It's also pretty much impossible to sneak up on them because of the magic "trip wire" that is still in T-Hunt.

The armor also slows me down a whole lot, but gives me no extra protection. I've seen terrorists do just as much damage to me with a P90 as with an AK47 when I'm wearing mainly assault armor. THEIR armor however works absolutely fine.

Did I mention how in extreme close-quarters I die in one hit from an SMG but it still takes me anywhere from 2-5 shots with a high-powered assault rifle to kill that person (who has less or no armor by the way)? Oh, and I often see terrorists point their guns at the ground, fire them, and then I get hit. That particular problem, however, has plagued every Rainbow game (at least for the consoles).

Just thought I'd add to your already extensive and accurate lists of problems

Anyone who claims it can't be reasonably done or too difficult never played FEAR. The AI in that game was the best I've ever seen, and there's no reason not to incorprate that kind of programming into a Terrors Hunt Scenario. Hell, even SWAT 4's AI was better than this!

quote:

Originally posted by Want2Snipe:
That would be what a game should be from a Dev that has pride on his work or a company that stands behind their product but, alas... we will never see that from the likes of UBI since all they are looking for is Milking that tom clancy cow (so to speak).

If Ubisoft had *any* pride .. we would have been informed of their intentions for fixing Vegas 2 by now.

The Ai behavior would be a step in the right direction!

quote:

Originally posted by noubiforyou:
Anyone who claims it can't be reasonably done or too difficult never played FEAR. The AI in that game was the best I've ever seen, and there's no reason not to incorprate that kind of programming into a Terrors Hunt Scenario. Hell, even SWAT 4's AI was better than this!

I know that Metal Gear : Snake eater is not FPS game but, I am really amazed with the tangos AI in this game and their realistic behavior during gameplay, they even contact each other for backup ... I really wish if FPS game like R6 Vegas 2 has the same AI, and yes, I am dreaming

sorry for bad english btw

Also, if they were to hold hostages...you should go about it a certain way before the timer goes off that counts down when the terrorists will execute the hostages. Stealth and using suppressors to silence your approach to keep the timer from coming up to tick down to hostage execution. The first one who messes up by forgetting to add suppressors will start the timer putting the team in rush mode to save the hostages. Of course you could go loud just to rush through and see how fast and efficient your team is under pressure but I'd suggest going stealthily!

EddieDeath

You just clearly described :

GROUND BREACH from BLACK FOOT STUDIOS

Here visit us at

http://www.blackfootstudios.com/forums/index.php?s=9146...6270e2d&showforum=13



cheerios

quote:

I know that Metal Gear : Snake eater is not FPS game but, I am really amazed with the tangos AI in this game and their realistic behavior during gameplay

Dude even MGS1 has better AI than Vegas 2, although i feel MGS2 would be the one to take from, room clearing is the shizz really.


wish we could have enemy hiding too with the cover mode we have in this game which is HEAVILY under-used by the AI!! Well if they used it correctly, at the moment they still run into an open space, spin on the spot a few times then get killed... hmm yes

totally agree with all the points made in the original post =]

quote:

Originally posted by EddieDeath:
I have this worrying feeling that the guys at Ubi think the current behaviour of the AI is "good enough" and doesnt need investment / development compared to other areas of the game.

If you play this game for the coop T-Hunt then hopefully you know what I mean. This is not about the spawn bugs but instead the AI as a whole. Today I found myself constantly thinking about what features I think the AI need and here are the distinctions I arrived at:


AI should primarly be attempting to achieve something i.e. holding hostages / stealing money / setting up explosive device etc. Everything they do should be directed to this end - as aposed to trying to kill the player team.

AI should primarily defend with a core team around the objective and fire teams patroling and scouting a perimeter. They dont give a sh_it and the player team aslong as they dont stop them achieving whatever the hell.

Once AI becomes aware of the player team presence it should become more alert and possible dynamically plan an ambush between the player team and the objective. Generally though it should only attack if it has a good chance and more frequently should fall back and camp.

So most of the time the player is attacking a well defended position using stealth / smoke / flash / grenade to give it that split second enough time to clear and progrss to the core objective.

AI should only start hunting and attacking if: they can first pin down the player team by covering all routes to the objective, and can then use smoke / flash / grenade to attack them once pinned down.

By attack I mean move towards the player team instead of laying in wait.

So effectively the player team has to constantly avoid getting pinned down, use stealth and evasion to get to the core objective and most of the time be attacking and entering rather than just camping and shooting AI as they aimlessly walk towards them.

Sorry I cant express this better but I really think its the core problem with the vegas AI and somehow I still hope that UBI will do something to make this change. I still believe rainbow six can be a truly great game and set a standerd for online coop if they stop shrugging off real AI coding and forget the gizmos.

This is probablywhat happened at Ubisoft after seeing your thread:

Community Developer forum reader (CDFR from now on) : "Hey Philip, we got a cool suggesiton!"

Philip Therein: "What is it?"
CDFR: "Lot of people think we should have better AI, you know, real AI."
Philip: "So?"
CDFR: "....So we should try doing that for the next game."
Philip: "*scoffs* What did you just say?"
CDFR: "..... We should try doing that for the next game."
Philip: "*angry this time* WHAT DID YOU JUST SAY?"
CDFR: ".... *cowers* We should... try doing that..."
Philip: "YOU'RE FIRED!"

Coding an AI like this would be a fascinating challenge. From individual actors to fireteams and ultimately an overall plan. Dynamically constructing a hierachy of plans and actions based on the primary goals and adapting these plans as the AI learns of player activity.

A key feature would be a replay (observer mode) ability so that you could see exactly what the AI did. Would be fun if this worked in coop so the entire team could rewatch a battle.

AI fireteams would communicate by radio so if you can take them out before they radio in it buys you time but a few spent seconds in the firefight and the entire AI knows your position...

What interests me the most is how an AI team could dynamically change its strategy i.e.

Move hostages to a different location on the fly if it detects you approaching from a given direction.

Pull back patrols to a tighter perimeter as its forces are weakened.

Change from smoke to tear gas or vice-versa if it discovered you have heat vision / gas masks...

What really grabs me though is if it would be possible for an AI to like I said before "pin down" and clear the player team - like a role reversal. It would have to attempt this if it had sufficient forces but be prepared to cancel the attempt and fall back to defense if the player team manages to get out of the pin.

Imagine Casino Vault - your team sets up on the vault - a few breach charges and flashbangs later and the vault is clear - everything looks good but one tango gets away. You hear the radio message but there is nothing you can do.

So you tell your buddies to hold position while you think up the teams next move - no sign of any more tangos...

Next thing you know - all the exits are blocked with supressing fire and c4. The AI dont expose themselves but can cut down any team mate who goes near a door.

Now you are pinned down and your only chance is to frag and clear an exit and try to get the team to a new position.

Given enough time the AI would setup fireteam(s) around the exits and attempt to frag / flash and clear the player team. If AI actually used flash / smoke / grenade etc to breach in - all from multiple entry points simultaneously - the player team would severaly pay for being pinned. The game would be turned on its head if players had to avoid AI being able to move towards them in force.

Its ironic that getting pinned down in the current game is the easiest way to win a map when it should be the primary thing the player team is attempting to avoid.

What if everytime you played a map with your buddies you didnt know what to expect. What if you weren't playing 70 tangos but just 17 but it was a hell of a lot tougher and unpredictable.

Maybe a lone scout will spot you and radio in your position before you can take him down or maybe a fire team patrol stroll passed your point man and you dont know whether to engage or not. Maybe you clear a room only to get pinned down and lose a man with still no idea where the primary objective is or maybe you turn a normally safe corner to discover the AI have dynamically setup a fixed machine gun.

The more I think about it the more exited I get because I can see a project like this would be incredibly fun to work on - somebody is bound to do it!