This is pretty exciting: XAITMENT

Someone on this forum suggested that XAITMENT might be used in VBS2. It's a bunch of products related to AI.

Check the website:

http://www.xaitment.com/

The more i read, the more i thought it was perfect for Arma

---------- Post added at 09:04 AM ---------- Previous post was at 08:17 AM ----------

http://www.youtube.com/watch?v=NAUIUNjuVHg

Very Hard to add into RV engine and it would be hell alot work to make it work without any flaws

It's been stated in a VBS2 roadmap, but it may not be licenced for Arma. The products are different.

BIS never really used third party middleware for Arma, but now they use PhysiX, so who knows... Animation, Lighting, simulations... there are plenty of supra cool techs available for licensing in your engine, but they come at a cost and you may be tied for the future or not be able to tweak them as much as you'd like...

Very Hard to add into RV engine and it would be hell alot work to make it work without any flaws

Why is that?

Good read here for fellow Xait-heads:

I really think that AI is the next thing that needs to be addressed. I guess I’ve said this a number of times; you’ve got these Xbox 360 and PS3 games that you play on your 60-inch plasma TV, and you’ve got tremendous surround sound, and the graphics look amazing, with everything running in HD, and the physics are amazing, and you’ve still got stupid characters

If you’re giving a character a sense of what’s in a room, for example, and a database of information about the contents, and giving the characters a logic tree that lets them make decisions about how they might use the objects in that room in something like a fire fight, then its about far more than just going to point-A to point-B. That’s what we’re working on right now.

Want Xaitment's baby nao!

Full interview here: http://www.develop-online.net/features/1487/Interview-Mike-Walsh

Way to dash my dream Dwarden! :(

Hehe, in all seriousness I wasn't expecting it to make it in to Arma3 but would have lobbied for it whether it was in VBS2 or not. At least on the surface by what they advertise -it just seems almost a too perfect fit for the last piece of the AI scheme that have always been missing in the series -that being intelligent interior AI, AI mesh, and seemingly unprecedented pathfinding ability. Now alongside a behaviour system allowing backround, culture templates (to factor into behaviour) and actually "thinking" about objects/guns etc..that are available for a needed situation -you can't blame a fan for trying :)

Nice that you cleared that up Dwarden.

Although I do understand the hype of ppl, seeing it on VBS 2, one would always wish it for Arma.

I guess the real problem here is the current AI path finding, if it was sufficient (in Arma 2) people would not even get interested in other type of middle ware solution.

Since it has been confirmed that Ai pathfinding will be improved, and there was even talk about 3D path finding for when inside the water, I guess we have nothing to be afraid, or want middeware solutions.

_neo_

Pathfinding is the main thing that's wrong with ArmA AI (and responsible for most of it's Darwin Award-grade stupidity), but hardly the only one though. I found their reactions to events around them really unrealistic, they're all calm even under the most intense fire, but also act somewhat slowly and IMHO, rather "robotic". Their senses, on the other hand, are way too sensitive. Then can see the player through bushes, from a huge distance and perfectly hit him using an ironsighted AK (once they turn, aim and wait a short while).

I think that it's another thing that should be looked into.

one thing, don't always believe that what's good for VBS is good for ARMA ...

different products, scope, aim, target audience, team size, budgets ...

yes sure there are tons of nice middle-wares ... but even if we wanted 2 best in each major area ... the game will be released around 2022 ...

so my suggestion is, expect less, be surprised later if there is more

Except AI still doesn't know how to navigate through buildings on its own. And can't pass through 2m wide holes in the fence. And can't step onto the 20cm high borders at Avgani (even though a human can do that without a problem). And can't properly drive. And and and

And so far a lot of stuff I've seen in VBS2 would only add to ArmA experience.

Then can see the player through bushes, from a huge distance and perfectly hit him using an ironsighted AK (once they turn, aim and wait a short while).

No they can't do any of this. If they did I wouldn't be the top1 shooter when playing with AI. Even with max skill and max precision they sometimes can't hit for shit.

And since when bushes make you invisible and stop bullets?

And so far a lot of stuff I've seen in VBS2 would only add to ArmA experience.

That is like saying [insert game] has feature [insert feature] that would only improve ArmA experience...what that has with A3 development beats me...

No they can't do any of this. If they did I wouldn't be the top1 shooter when playing with AI. Even with max skill and max precision they sometimes can't hit for shit.

give yourself a cookie...

one thing, don't always believe that what's good for VBS is good for ARMA ...

different products, scope, aim, target audience, team size, budgets ...

yes sure there are tons of nice middle-wares ... but even if we wanted 2 best in each major area ... the game will be released around 2022 ...

so my suggestion is, expect less, be surprised later if there is more

Still though, if you had a small team to work on integrating ExAItment into Arma's version of the RV engine as a side projeect that would span several Arma's lifecycles, we could probably have it ready for Arma 5/6 no? I don't see why not, the budget could probably be covered then as arma's 3, and 4 no doubt will make BIS alot of money, and the new arma w/ exAItment will make even more due to better AI! Do you not think this is reasonable?

one thing, don't always believe that what's good for VBS is good for ARMA ...

different products, scope, aim, target audience, team size, budgets ...

yes sure there are tons of nice middle-wares ... but even if we wanted 2 best in each major area ... the game will be released around 2022 ...

so my suggestion is, expect less, be surprised later if there is more

:( You made me a sad arma player

I think the AI pathfinding, be it on foot or in vehicels, is one of the biggest flaws of Arma and widely criticized in Reviews of Arma 2 and OA contributing to the bad ratings. Therefore I think that Pathfinding should be improved a lot!

Frankly I really don´t care if this is done by Xaitment or by a BIS solution. You guys are very creative when it comes to solving problems and developing improvements.

Since Xaitment seems a bit to expensive and time consuming to implement I hope for a much better pathfinding made by you guys.

Make me a happy Arma 3 player :)

I think the AI pathfinding, be it on foot or in vehicels, is one of the biggest flaws of Arma and widely criticized in Reviews of Arma 2 and OA contributing to the bad ratings. Therefore I think that Pathfinding should be improved a lot!

Indeed, particularly inside buildings.

That is like saying [insert game] has feature [insert feature] that would only improve ArmA experience...what that has with A3 development beats me...

Thats not so much a leap when both games are based off of the same lineage.

I took a look at the website.

Did not convince me.

I just think the VBS team lacks in-house AI expert.

This can be made in-house with less effort than you think.

They did not invent the wheel, just took a handbook on Game Theory and implemented it.

We do not know how it will work in VBS and what the hardware requirements will be for running a server. They might end up running the AI on separate machine for example, if they want deep AI for every single soldier.

ArmA2 already used databases for threats and weapons and determining when and using what to attack. ArmA2 had also quite sophisticated system for using terrain as cover when approaching the player.

The Pathfinding was a problem mainly because of grid size and battle scale - Bi did not want your CPU to go ablaze.

I am trying to think about things AI is missing in ArmA and quite frankly it's all tweaking or little things like swapping weapons or using vehicles in conjunction with infantry properly.

Hmm, in-house AI expert? If you ask me, BI are the leading edge AI experts at least for this type of open world combat sim. Who else has anything close in terms of amount of unscripted AI that, short of urban cqb, fight and move pretty effectively? No one does. Most developers in the military genre these days either corral you into their movie labyrinth of dressed up corridors or just outright omit AI in favor of MP only.

I honestly can't think of any other game in any genre that comes close to this. GTA? They're good chasing one person around a cityscape that omits indoor pretty much entirely. Oblivion games are similar. My beloved M&B has some of the best large scale CQB ever (up to 1000+ units) -all taking place in a "closed room" for battle -not the same as everything taking place on the world map.

BI has nailed the large scale combined arms AI up to everything but CQB and interiors. Implementing believable interior and urban CQB (using all objects including vehicles as cover, clearing rooms and fighting from doors and windows) -is not a matter of tweaking. It's either got to be built from ground up or use something like Xaitment offers. Thats not to say BI in-house is not building this version from ground up, but it certainly ain't just a few tweaks.

it certainly ain't just a few tweaks.

Why? The A* search algorithm looks the same in all games. The difference is in graph generation which is done manually or based on mesh in corridor shooters or by using terrain mesh or auto-generated nodes in case of open terrain.

The density of grid was increased in ArmA2 and I think the results speak for themselves.

Shooting out of windows and doors ironically IS a matter of tweaks. Ability to do it is already in the game, the AI just does not do it when you'd think they should. The buildings have AI nodes and pathways in ArmA games.

Add them as good cover spots and they will use them.

Finding that hard to believe :)

Obviously, VBS2 hired them for a reason -whether that be Xaitment takes off a major workload beyond the tweaking category or that like they advertise -they specialize in behavior with pathfinding.

If Arma could have added window (which are not there own objects afaik) and doorway leaning/firing via tweaks -they would have done it by now. I've never seen vanilla AI take up a new behaviour via reaching a travelway node. Arma 2 windows were worse (then OA) with 2 enemy AI facing each other face to face not taking up any sort of appropriate behaviour -these are things Xaitment at least advertises on and I'm assuming with good merit as again they were hired for the job.

Finding that hard to believe :)

Obviously, VBS2 hired them for a reason

Running the risk of sounding patronizing I'll just say neither of us knows the reason BIA hired them. Could be they need the system in ASAP, could be they do not have capable programmer on their team, could be they were just offered very good deal.

It might also be the AI in VBS has some extra functionality like setting up roadblocks or moving objects around or reacting to voice commands or typed commands that VBS2 could not handle but that is not needed in a video game.

Special requirements.

Whatever that is we have AI leaning out from behind corners of buildings why do you insist using door and window frames is such a leap in technology? Do you know how game AIs are build?

And since when bushes make you invisible and stop bullets?

When you're wearing camo, they make you difficult to see and obscure precise aiming. This doesn't seem to matter for AI.

If Arma could have added window (which are not there own objects afaik) and doorway leaning/firing via tweaks -they would have done it by now.

IIRC, they're technically in-game. They just don't work, due to AI never "deciding" to use them.

When you're wearing camo, they make you difficult to see and obscure precise aiming. This doesn't seem to matter for AI.

So a gunbarrel smoke (or flash when it's darker), a sound of you shooting from behind a bush should stop AI from sending a few bullets there?

Bushes don't stop me from doing the same, why should they stop AI?

You can also search for ArmA2 mythbusters thread where there were some tests of AI and bushes and all showed that AI doesn't see you through them. In fact you can even hide in them after engaging AI and it may stop firing at you (in fact that's often a case in my games which is annoying since I won't stop shooting the moment AI runs behind the bush)

In one of my games I also witnessed a ridiculous moment where two enemy AIs were facing each other and standing 5 meters apart but not engaging... because there was a bush in between. Even though I could clearly see feet of the enemy behind that bush.

If that isn't enough of a proof to you - open the editor and put yourself in an AI controlled squad in some bushy area. When AI will order you to shoot someone you will see a red reticle pointed at the enemy. That's exactly where AI believes an enemy to be. And the moment an enemy disappears behind a bush the reticle keeps moving with the same speed and will keep moving even over an empty desert. The AI will notice an enemy only when it will appear again or will make too much noise.

neither of us knows the reason BIA hired them

It might also be the AI in VBS has some extra functionality like setting up roadblocks or moving objects around or reacting to voice commands or typed commands that VBS2 could not handle but that is not needed in a video game.

We do know that VBS2 is showcasing Xaitment specifically for high precision pathfinding including scaling of high buildings

MsgCLX84RZo

Whatever that is we have AI leaning out from behind corners of buildings why do you insist using door and window frames is such a leap in technology? Do you know how game AIs are build?

Leaning was and is a great addition to the systems AI. That said, there has been little progress in getting the AI to do anything differently once they are inside a building -they still fight as if outside minus the lean.

I'm not here to argue or cock waggle over who knows more about the building of AI as I'm sure technically you do. That said, degree in AI withstanding, if Xaitment could provide (or BIS) the type of AI they espouse they can, regardless of reason for hiring them -I'd prefer that'd it'd make it in than not.

Bushes aren't the problem, it is long long grass. Turn grass off on a mission with AI and you will blitz through, even on max precisiona nd skill, turn it on to full grass and you will get owned.

Running the risk of sounding patronizing I'll just say neither of us knows the reason BIA hired them. Could be they need the system in ASAP, could be they do not have capable programmer on their team, could be they were just offered very good deal.

It might also be the AI in VBS has some extra functionality like setting up roadblocks or moving objects around or reacting to voice commands or typed commands that VBS2 could not handle but that is not needed in a video game.

Special requirements.

Whatever that is we have AI leaning out from behind corners of buildings why do you insist using door and window frames is such a leap in technology? Do you know how game AIs are build?

Its pretty clear why they were hired (and thats probably the wrong word to use anyways). Purchasing that tech allows them to take an existing well rounded product and drop it into their engine. In the short term its cheaper than writing their own system that does the same things. Its a pretty common occurrence in any software development cycle.

Look at all the libraries that go into modern games that are not made by the studios themselves... You have FMOD and Miles for sound engines, you have Havok and Physx and Bullet for physics, and the list goes on.