Star Citizen Live: AI Roundtable Q&A

20 May 202201:06:581,937 linesWatch on YouTube ↗

Transcript — every row opens the video at that moment

00:00:02
Star Citizen Live, the AI round table Q&A starting Jared Huckaby and members of the AI content feature and tech teams. Yeah. Hey everybody. Welcome to Star Citizen Live. AI round table Q&A, I'm your host Jared Huckaby. Welcome to my my home in beautiful sunny Manchester, United Kingdom. I say that and people think I'm like, "Oh, yeah, I know that's sunlight." I am lit by only sunlight right now. It's actually I've been here for 27 28 days now and it's only rained two of the days we're here. Everybody was worrying me about about the weather here and it's actually been really nice. So, I'm going to say that now and then it's going to be a week of storming. Joining us on the show on the show this week members of our esteemed AI content tech and feature teams. We've got Marius.
00:00:54
I'm Marius. So, I'm yeah, I lead programmer. Um I'm for a long period here with in the Frankfurt studios. I think it's more than around 6 years. Um Nowadays I'm taking care of planetary navigation tech. So, basically generating navigation mesh at runtime around the players vehicles. And yeah, this tech is kind of close to my next priority is getting into the work for the navigation links. And a bit of to explaining what navigation link is are Um it's about NPC being able to to use air locks ledge grabs.
00:01:44
Ladders, transit like elevators and also metro station all that. So, yeah. This is what's going next. Now, let's not understate undersell that planetary nav mesh. That is a huge addition to the Star Citizen. That is what's going to allow us to put AI out at all of these outposts and points of interest so that when players go and explore these down wrecks and everything, uh they're no longer just going to find an empty abandoned location. AI can you know, once we get this planetary nav mesh AI is going to be at these at the derelicts, the the outposts, the the caves, you know, wherever. It's It's It's It's an exciting foundational tech. Exactly. Exactly. Um I just want your mom to be proud of you
00:02:31
when she watches this. Thank you. We are proud of you, Marius, as well. Uh Andrew, who are you? What do you do for Star Citizen? Hello. Uh my name's Andrew Davis. Um I'm the product owner of the um AI feature team. Um that's a very uh kind of grand title but basically what that means is I take all of the uh crazy mad designs of our designers and and all of the other requirements of the other teams and then I passel it up and turn it into reality um with lots of very boring uh planning meetings but at really as at the end of the day what I'm doing is taking their kind of dreams and turning them to little bits of work and passing
00:03:19
them on to our really talented uh programmers and and animators and uh making those work uh hopefully. Um and I'm also uh senior developer and I've been working on uh getting our uh chow line um AI up to scratch. And this is something I worked on before but it's kind of uh lapsed a little bit but we're uh we're kind of getting it running again so I can pass it along to the content team. And the chow line is a uh a system that allows is a social bit of a social AI that allows our guys to go and grab a tray from the mess hall, grab some crockery, and then go and use lots of different
00:04:07
machines that each produce bits of food and drink, and then go and sit down at tables and eat them and drink them and do that to their heart's content. Um and it's a system that has quite lots of little different bits working in a way that not if I if I'd worked on a previous game, you probably will have that all done in one canned animation. But because this is start of the season and we like to make things as we like to go that extra level with lots of detail, each of these machines is a separate thing and that means there's lots of places where things can go wrong like you don't get the the the crockery that you want or something else or let's say there's a whilst your your getting your food, suddenly there's an alert over the tannoy and all of
00:04:56
those characters have to drop what they're doing and leave their plates and trays and go off and man their battle stations. So it's dealing with all of those things and like how do you come back to that and deal with that. So that's kind of taking lots of stuff that we've already worked on and then putting it all together to create a like a really complex system and with all the complex details all the bugs that come with that. Um so that's basically what I do. And and a lot of your focus these days is on Squadron 42, but my most much of that work will make its way into the PU once it's you know, proven out and stuff. So Yeah. Uh Fran no stranger to Star Citizen live. Who are you and what do you do?
00:05:43
So I'm Francesco Roccucci and I'm the AI director of the teams, so the AI team. Mostly my responsibility is overview of the implementation and all technical implementation of all the elements of the AI, so making sure they're implemented correctly, that they're up to the standard, you know, working with the design to look at the logic or with design and animators to look at features that they need, how do we make them, you know, and then making sure that they're dispatched to the right person or some stuff I do it myself because I really like it. So, of course, like, you know, there are things that I've been working on recently like the dynamic conversation that I hope then will go on, you know, soon. It's something that you would add a lot of life to a lot of the behaviors we're working on. Um but, you know, there are stuff like I
00:06:30
know I support a bit of the different features in general like the usable factor one to do that are like improvements especially on the tool side and you know, and how we debug stuff and how efficiently we are creating things. Want to make it much more efficient. And then a bit of navigation stuff with Marisa I try to support whenever I can, you know, different parts. That conversation stuff is cool. We actually showed it recently on a on a Star Citizen Live. Um just I I think everybody who's ever played a video game has has had those moments when that when when when the cutscene isn't fully rendered when the cutscene isn't pre-rendered when you know, when it when it's actually being rendered live in the thing which is always needs need appreciated and you're trying to talk to somebody and then somebody's big ass head comes in front of you and this is
00:07:19
as they're trying to get around you while you're talking. You're like Christ almighty, let's go. It's like that. So, it's it's it's little things like that you know, it little things like that build up and make the the immersive universe that we're that we're that we're trying to get to. Yeah, one thing one thing that we really uh we're working on in the animation was recently I did is like making sure that when people are in conversation, you know, they create some area of avoidance, right? Avoid those sort of things and all this polishing that, you know, to achieve the the the great quality we want. All right. Uh Stewart, first time on on Stars and Life, who are you and what do you do? Uh hello. I'm Stewart Chard. I'm an animator here at Cloud Imperium and um you know, on on Star Citizen what I do is try and create life
00:08:08
through NPCs, basically. Um bring characters into the environments, into the universe, and um get them to do performances that just add that extra sense of um emotion. Um so I would say characters, but like there's a lot of NPCs. So um yeah, basically adding life. That's a shortish way to explain it. Um Some folks might not understand why an animator is on an AI team. It would seem it would Isn't there an animation team? Why are there animators on AI teams? Um one one reason is because um like well, one thing that I work on a lot as well is these things called
00:08:55
usables. And um people A usable is a thing that somebody can interact with, uh like a coffee machine. So I'll uh go out there and do the research on how to use a coffee machine, and put together sort of like a flow. Like Luckily, I drink coffee a lot. Just using that as an example. Uh but I'll go out and video tape myself using my coffee machine and put it together as a uh what people would call like a state machine, or some people may refer to as a flow graph, and it would basically explain that uh a flow of how to use that usable, the coffee machine. Um and by doing that we maintain like a sense of realism, um performance, and just adding more to that expanding universe. An AI that can
00:09:43
knows when to trigger what animation when based on what it's trying to do. Yeah, for example, bartender knowing on how to make a highball, or knowing how to make a like how to get a small glass, knowing how to make you know, all the all the different drinks, you know, um, so yeah. All of those different things. They all require different um, performances. And so, I'm in charge of making sure that performance is authentic. Got you. And that's why animation is such an important aspect of of the of the VR work. Yeah, you had something Yeah, I think you said you had something you wanted to show us real quick before we jump into questions and stuff like Uh, one example of what I do is actually go around the universe um, through these like awesome environments that people have put together. And I like pose characters that are sort of interacting with different usables.
00:10:31
Um, and then I speak to our animation director, Steve Bender, and request for mocap shoots to be done. Um, and I also talk to I talked to Fran and Dan and multiple designers about um, how we can actually get this to work. Um, and something that's also been brought up in some chats lately are commuters on train stations and things like that. So, I thought I might show some of the poses and then the shoots. Um, there's some work in progress. Drumming is obviously what I like as well. So, I'm going to put that in. Um, but this is something as that I'm really interested in adding. Just some characters with different kinds of postures, things like that. Okay. So, yeah. And this is recent stuff. This was just done a a week couple weeks ago? Yeah, I believe it was done Tuesday.
00:11:20
Oh, yeah. It's super fresh. Super fresh. So, it's straight The last one where he fell asleep, that wasn't actually you didn't ask for that. He just You you overworked him so hard that he had he just fell asleep. on the air uh uh folks who have read the letter from the chairman that that came out just this week um in the section about the studio and the new studios that we're building uh there was a small blurb about a brand new 4,500 sq ft performance capture stage uh that's that's that's being uh put together uh right across the walkway from our new office. It's our our our new office in the UK is a is a campus. It it it's part of a campus. We have the top three floors of the Manchester Goods Yard building uh which is the one that you see in ISC hosting and everything and we're going to go inside we're going
00:12:07
to go inside uh starting next week. Um and then like us you could throw a rock and hit it like right next door across the walkway uh there there's another building with it where we're building this giant performance capture uh stage uh where we'll be able to do uh uh pickups like we just did like Stewart just showed you for for uh just anytime we need to add more NPCs into into places and make make the place feel more alive uh to do pickups for Squadron 42 things that we you know realize oh we need a bridge for this or that or whatever or even when it comes time to it production for you know additionals you know sequels and stuff for Squadron 42 and stuff like that. So uh I recently Yeah, I recently got to actually see it as it's being built and it is so exciting. It's huge.
00:12:55
And there's actually a picture of it on the letter for the chairman chairman when when it mentions it there's a link. Uh uh if you click the link you'll see a big picture of it and screen state and then there's a a third uh building kind of catty-corner. It's it's weird. It's like an L and that's where my new stage is going to be where we're going to be where we'll be doing these shows uh live and in person and stuff like that when the time Do you have to wear the motion capture suit for that as well? No. No. We're we're we're we're we're we're we're we're we're finally for the first time in the 8 years I've been here we're actually going to have like a proper set. So so they they they've they've That's got to be fun. I've paid my dues. I've paid my dues and and enough with shelves that I've stolen from other places in the building and put my own personal stuff. when you when you have a nice window with the sunlight coming into the room like right
00:13:43
you have right now. That that would be that would be spoilers. That would be spoilers. It's going to have a window. We'll leave it at that. That's cool. That's cool. Uh Dan, last but certainly not least, who are you and what do you do for Star Citizen? Hello. I'm Dan Bicker. I'm a senior designer on the AI content team. And as a designer, basically, we're kind of the guys that are the glue between all of the other teams. We tend to be the ones that um you know, we'll we'll get a request in, say for something I don't know, say something simple like a seat, and then we'll go, "Right, okay." Straight to the programmers, "This is the thing that we want. Here's the list of requirements." We we try and figure out all the requirements that might have not have already been thought about. And then leave them to mull that over that for a little bit. Might be a meeting or two required for that. And we'll go to the guys like the animators and we'll say,
00:14:30
"Hey, we need these animations of this this guy sitting down on a chair." And then we'll go to the artist and we'll say, "We need a chair." You know, and we'll just try and pull all of these different things together, and then usually, you know, we'll we'll get all of this content back and we'll try and put it all together and more often than not it doesn't work. So, we'll go back to the the programmers and we'll say, "It doesn't work." We'll go back to the animators and say, "It doesn't work." We'll go to the props artist and say, "You've missed something." You know, so you keep going back and forth all the time just trying to figure it all out and see where the problems are and hopefully by the end of it you get something that does eventually work. Uh that's kind of it in a nutshell, really. But I do actually have a demo of one of our more complicated usables because one of the main things I work on in particular, I mean, I'm you know, in AI there's just so so many
00:15:19
different things that are touches, you know, like like like all of the usable that you know, most of the usable that a player uses, you know, we'll want 10 PCs using them. there's a lot of tech involved in in AI as well. Like one of the most recent things I've been working on is like AI reactions. So making sure that they're reacting to certain things. You know, so if they're getting shot, need them to react to that. If they're in a usable and they get shot, then you need to make sure that there's a system in place to make sure that they react accordingly to that. And we're also just starting with the like the low and the medium type of reactions as well. Like, you know, listening to a terminal announcement, things like that. The kinds of things that we we want them to be able to you know, be aware of as well. And also I mean they could go on forever. It's like things like dynamic conversations and things as well that
00:16:06
have already been mentioned in wildlines. But I do have a demo here of one of our more complicated usables or complex usables. It's um So here we have a wall panel, which is essentially it's this is a full full-blown usable. And in here this is this is what houses the Oh, goes in there. There we go. So the the wall panel is the thing that houses Okay, he's going to walk off. Typical. It's it houses the ship components. So things like shield generators, quantum drives, relays, whatever it ends up being. And inside there you're going to have things like fuses or whatever other kinds of subcomponents
00:16:56
might be relative to that that component. And they're going to affect the statistics of the the component in some way. Most likely just like, you know, certain fuses um broken, you would probably expect the thing to not work as well as if it had three fully functioning fuses. Mhm. So the player is going to be able to, you know, go in here and do whatever they they like with the fuses they could swap out themselves if they want to. And um you know, place a new one back if they need. Say that one was broken. Close it. You could open the the next one. Same kind of thing. And this will all play into resource gameplay and engineering and stuff which we'll be discussing
00:17:42
Exactly. Yeah. Later this year. Interestingly, that doesn't want to close. That's a new bug. Typical. So, yeah, the point is though that the NPCs, we want them to be able to do all of this kind of stuff as well. So, in the future when you can hire a crew, for example, you'd want them to be able to go up and change these fuses for you, you know, without you having to do it all yourself. Right. And while what we've done so far is very squadron focused, we've also obviously had the PU in mind as well for for just making sure that anything that we do it will continue to work there as well. So, actually let's um just going to reload this real quick. It's always like we do a live demo. It's always like we always do live. Like
00:18:31
I don't I don't know why. say that one good thing is that in this case your behavior are done with those handle uh wall panel and already open or or you know, in different states because the players will leave them in the wrong states. Yeah. Yeah. Yeah. So, there's there's logic in that to say well, this door's already open. So, you could you could tell the NPC to open the the wall panel door and they'll still check when they get there. You know, it it it am I am I it must be the new people who think we do pre-recorded things cuz the people who have been here for 10 years, they know they have to know by now. We always do live. It it warts and all. I'll be honest though, I had a significantly worse bug like half an hour before we started to call. Managed to get that fixed though. So yeah, this this is an example of an
00:19:21
NPC actually doing something relatively new that we've we've developed which is assignments which is this this assignment in particular is it basically just tells this NPC to find this wall panel and just use it forever because there's going to be plenty of situations where you know, we want in in Squadron or wherever, you know, we're going to want these NPCs just standing in that place because we know that as the the player's going to walk through there, that it's going to look good if somebody's there. You know, so so this is just a behavior now where the NPC is going to just keep going through all of the different actions that they can do on it. So they can do like inspects on the component, they can do inspects on the fuses, takes, places. Um if they've got like a flashlight or a data pad, they could use those as well. Um cuz that's another thing as well so
00:20:08
that the NPC is aware of the things that they have on them. So the equipment that they have available to them. So if they have like a flashlight, then they'll use that. If they don't, then then they're not going to bother trying to use use that action. Um but one of the really cool things about this is is how the fuse itself is a usable, the component is a usable, and the wall panel is usable as well. So you've got like a three-step hierarchy of usables and the NPC is still able to search for the fuse, interact with the fuse directly, and understand that it's got the component and the wall panel between it and the fuse and still act accordingly. So so ba- so basically what you're saying here so it's like I said, you got you got the NPC comes up here, it can recognize that there's a door, it opens
00:20:57
the door, it recognizes that there is a giant component inside here, what component it is, what it's supposed to do, and then it recognizes that even within that component there are sub components and they all have particular jobs and purposes and it can sit there and it can go through and and validate each one. Exactly, yeah. And we can also put tags on the different fuses as well. Say Say one of the fuses was broken, we can tag that up as like damaged or broken or whatever it might be and then we can give the engineer a task to say find any broken fuses and then go straight to those ones and replace them with with fresh ones. Or upgrade them to a different different thing. Could be. I think that one one important thing is that these assignments that Dan is talking about are sort of API, right?
00:21:44
So, at the same things, I mean, connecting what you were saying about the hiring, right? If you hire an engineer, you might have an API of his behavior of what he exposes to you. That is the same thing we expose to level designer or mission designer, right? It's just that the mission designer might want to loop it forever on something because of of, you know, visual reasons, but you as a player might just say, "Can you repair this panel or can you inspect this panel to verify the wiring is correct?" Or maybe, you know, you have an alert in your you know, main board computer and it says like, "Oh, this fuse is broken." Then you can send one of your crew member to repair exactly that fuse. Those are the API that Dan and the other designers can expose from from their behaviors. And I think with the custom assignment that we literally implemented, I think this there's also these extra things where
00:22:31
it doesn't require engineering job to do this type of things. Then we can really expose better, you know, functionality that people can use. Now, well, before we move away from this, Dan, I have to ask you, what's going on with the pipes in the left there? Is that blood? I think it's paint. It's paint. And paint and rust. You got you you got to You you you you you got to keep better care of your Idris. Yeah, I think I'll have to make an assignment for cleaning them. Yep. Are you sure it's not the blood pipes coming from the medical bay? And for those of you who are wondering go there and just see where we are in the ship, but I'm not allowed. And for those of you who are wondering, yes, that was an Idris. That that that was an Idris. All right, you can go and stop this here. All right, so that is that is our that is our setup to
00:23:20
to to trans to to transfer over into the Q&A portion of our show here. we wanted to show that and everything. It it's it's it's it's a bit about how NPCs work throughout the ships and stuff like this. It's currently being developed you know, for Squadron 42 primarily. When the systems are all proven out there like most other Squadron 42 systems well, they'll be converted and brought over to the PU. We say this because the number one most voted up question in the thread this week and by a wide margin and then repeated at number four and repeated at number eight and repeated at 10 and 15 whatever was how will hiring NPCs work? Um we're not going to get into how the hiring of them works because it's actually not an AI test. That's that's actually a game design task. That's not
00:24:08
the team that we have here. But we will talk a little bit about the process of making those NPC crew work and and and what's going to be involved in them and stuff like that such as we just showed. So um we'll start right off here. What is the development path for hireable NPC crew implementation? Like what are the design challenges, the prerequisite technologies, the known roadblocks, the gate posts, everything besides just a lot of time? Um what are we talking about here? Where do you want to start? Okay, I'll start. I think like so some of the stuff I think that Dan just showed I think it is part of that, all right? I mean, Dan you'll be involved with a lot of those behaviors that just engineer that that is the same, you know, where we have um
00:24:58
characters that will take care of your ship, you know, or you know, help you to land or repair, you know, try to refuel your ship and stuff like this. So, all those behaviors are the basics for, you know, first you need to do those behaviors and then you're going to hire them to do that behavior, of course. Uh one part is the assignment and the custom assignment work that we were discussing, that is the way in which, you know, let's assume the player hired three guys first of all, right? So, it means that they are assigned to you, they're part of your crew. Because that way when you take that ship, you know, let's say they are there already, first of all. And then you can give them commands. And the way you can give them commands in essence is despite the UI, you know, the player facing feature, you know, works, it will eventually turn into the request for the assignment to the AI. And the request for an assignment to the AI means that basically we have something that we call master craft that is kind
00:25:46
of a that a a same machine on top of the regular activities and some activities that assumption describes. And and this is where the API and the custom assignment comes into play, right? Because we have a custom assignment activity that just can take whatever task you want them to do as custom as you want and they will just do it. And then there are things that we want the activities to expose. It's like I'm not an engineer, but if you ask me to repair something and you hire me, right? And I'm a janitor, I can try. Now, I might not be that good, you know, I might it might take three times longer than what an engineer would, you know, would take. And possibly I would put a fuse that is also broken because I cannot understand if the fuse is broken or or or or working. But I think we want basically NPCs to be able to perform job, you know, very likely. Maybe
00:26:34
poorly. You know, and that will it might affect their their morale, right? Because maybe if you hire somebody as an engineer and you give them janitor duties, then maybe, uh you know, you will have like some morale, you know, going down in respect to the type of of task that you give. So, all those things are already supported by the systems, right? So, some parts are just the way we expose them to the different behaviors. And yeah, and then a lot of this is polishing, right? Because I think people want us to have a good experience on the animation side, making sure as Don said that you know, this usable stuff correctly, props are set up correctly, right? That you have the different type of fuses and and stuff like this. And then I think another part from my perspective that is important is how
00:27:21
this crew is managed by you, right? And how persistence, so the persistence work we are currently working on as a company, right? It will be fundamental part because if I hire somebody, server because down, comes back, well, I should have the same crew that I just paid. You know, and I should be more or less where I left it, right? And trying to do the same things that I left them to do. And then possibly the simulation plays because if I leave, let's say, my engineer to do a repair, and I leave the hangar and they're repairing or my deck crew, let's say. I I say like, you know, can you repair this ship? And then I go with my second crew on a mission. When I come back, it's not a big they're not active, then the ship should not be repaired already, right? The time has passed. I expect that to still happen. So, those things are more or less the type of challenges and the core
00:28:09
functionality that we we need, you know, for the full-blown experience, of course. But a lot of the high fidelity, I would say, you know, experience in terms of, you know, abilities of doing those behaviors are quite there. Of course, if the loop is not there, the gameplay loop, let's say, repairing, if it's not there, then of course, we don't want to fake it on the AI side, right? So, we just prepare the stuff that then it will interact with the fuses. What Don was saying, right? Right now, we could tag the fuse to be broken. But once there is the gameplay loop in which, for example, using a lot of a fuse might cause that to break, and then the game code behind the scene already tags it or maybe just an L component we can use. Using that data is trivial for us because all the rest is already there. I think you know it's
00:28:59
this is basically our process right now. If anybody wants to add some more. one of the one of the things that is a big challenge with all of that and and with all of the social AI behaviors that we do is just knowing because we've got these very complex systems just being aware of all the different ways in which things can fail and then having some good behavior to fall back. As you said if in that that the example like if the the ball panel's already been left open that you can cope with that and and that's something that other games don't necessarily have to deal with because they they very tightly control the the situations in which that AI is working. But for example one example sorry Jared if you can just just little thing. What
00:29:45
Dan showed I think is really cool. If as a player he was going to remove one fuse the AI inter you know interacts with the fuse so he knows that there's no fuse there. You don't risk the the AI would choose to play an animation on the missing fuse because there is just nothing to interact with right. So I think it is more complex but then eventually what you get as a you know as a as a as a repaid of your hard work is the fact that this really works systemically. Right and and you know the way you configure that as Dan said if you don't have the flashlight you just don't use it. You're not playing the animation of the flashlight without the flashlight. And also leads to other behaviors like I was thinking about in the chow line and and the the mess hall if there's lots of people having to go off or the player leaving loads of trays
00:30:35
behind then there's going to have to be a janitor character who comes in and picks up those trays and takes them takes them back um so we don't get a a a a a mess hall that's just full with with trays. And as Dan and Stewart showed, I I I imagine animation is going to be a major factor in this Absolutely. because there are metrics, you know, we we we have metrics in sure and and many many ships follow many of the metrics, but there are always going to be these outliers and these it's you've got you know, nearly we're nearing 200 ships at this point. So, you know, you have to go through and prove all these NPC crew out over every single ship with an interior is is going to be a tremendous amount of work. Yeah, like I was I was going to just say
00:31:22
as well like for for my perspective like a lot of the challenges are just with dependencies. So, if you have a task that the NPC needs to do which involves like needing a a multi-tool so that they can weld or repair or whatever it is, they need to be able to get a hold of that thing. You know, so they need to have lots of ways of getting a hold of that thing as well be a locker or you know, equipment chest or whatever it ends up being or even just a multi-tool lying on the ground. They need to be aware of like oh that thing's over there. I'm going to use that instead. But like you just said about um animations as well. Uh we have that all over the place. I mean, one of the things I'm working on right now is the deck crew. And that's NPCs repairing ships like the Gladius, you know, I mean, a lot of what we're doing right now is very squadron focused so that the player can see these
00:32:10
NPCs repairing the ship physically. But if you went into a huge amount of fidelity on every single possible area of the ship, it would just be a colossal amount of animations. And you know, even just things like say say you have something on top of the Gladius, the NPC needs to be able to get up there. You know, which means that they need to understand that in order to get up there, I need to grab the ladders, move them over to the gladius, and then go The ladders are the ladders are trolley, so need to be able to use the trolley. to be able to push them, move them around, and and then when they get up there, you need to make sure that they've got the animations to enter, and it's all the correct metrics, which isn't always the case. You know, so then you got to go back to the artist and say, "This isn't a metric." And usually you
00:32:57
have to design a metric first, and then get them to change their their baby into something that you've you've written down months or sometimes years later. Yeah, I think the challenge is also to give them them the tools to do this more as efficient as possible, right? Because I think you guys might have also a way to override all of these with a custom animations, for example, for each ship, but I think the critical bit is how quick you can do that, right? And I think how quick Stuart can go through in a test map, and this is why we have done like a lot of also work on the custom assignment, because we can use them in the future test, and we use them to try to speed up a bit this process, because Stuart needs to be able to go there and say, like, "Play the I you know, going up the ladder animation." You know, without spending hours on trying to reproduce it, and the same for them to bring that back.
00:33:43
I don't I don't want to keep making AI crew sound like a nightmare absolute nightmare scenario, but we do want to effectively communicate what an initiative what it what what a endeavor it truly is. Uh to what Dan and Andrew said as well, those resource you know, the fuses and everything, those are all part of the whole resource management system, and that system has to be developed and proven out and usable by players first before you then figure out how to how to expose all that stuff for AIs, too. So, this is my this this is my way of saying it I don't I don't we're going to move on from this topic here, cuz there's a lot of topics we wanted to do, but yeah, it's I just it's important to I want folks to under to to just be sure that they understand uh to properly manage uh We are shooting for the moon with this
00:34:33
thing. And and and and when when it is complete and when it is uh fully developed, it'll be a thing unlike any others. But it's also starting at the process, right? So, we get, you know, the first step, then you you can iterate, you can make it better, you can add that and more fidelity, the more you can, basically. But NPC crew is not a soon thing. It it it it it is a after many many other things are are done. On this note, uh sorry to to also say you one of my points. Uh the idea is that we also try to uh reuse what we we already So, basically, the the social AI, we already going to implement behaviors, activities for repairing, right? The the when you hire hire them, you just try to use the same something to use the same we only with the input dynamically from the player because you want to repair your your
00:35:20
your ship. You want to actually work for you. You want So, basically, you just give guidance on already existing behaviors that we are we already trying to implement. So, at least not to duplicate and not to put more work on this. I think Jerry, now you know that you need to stop us talking because we have like too much stuff that we want to say. Mario, if you're working on the the dynamic nav mesh right now, wh- wh- when that when that's in when that's in uh um uh uh later this year, uh it'll allow people to go to these uh certain POIs, you know, not every POI, it's up to designer where they're going to use it and everything. Uh but they'll be able to go to certain POIs, certain things that we're going to show here in ISC real real soon in the next couple weeks,
00:36:09
and encounter NPCs, friendly, hostile, whatnot. For the hostile ones, how's it going to work? Dynamic isn't the entire planet. It that that's the whole point of it. It's it's it's pockets of the planet where the players are anything. Um what's what's kiting an AI going to be like? Are we going to be able to to to to go to an to go to a a POI and kite somebody all the way back to Loreville here to to to the main landing zone or something? What what's What are we talking about here as far as scope? In theory, yes. Yes. So, the we already kind of showed a bit during the CitizenCon, right? The the outpost and the NPC going moving around trying to I don't to uh selling stuff, having combat with
00:36:57
you. At that moment actually the the planetary navigation, it was close to end, but it wasn't finished. So, we weren't able to use it, but now we we we should be able to to use the the planetary navigation in this kind of similar scenario. And I think mission teams are trying to create some some missions around this. Uh and yes, in theory as I was saying uh because we We we got we got same answer. I can say that we handed it over to design that that already working on that because I think, you know, that I think it's a cool thing that right now they are already started using it and they already, you know, they can provide us feedback already, right? It's it's it's internal development right now and as Jared said, it's something that, you know, people can experience in the this year at the end of the year.
00:37:45
And because you said that if we could move with NPC everywhere, even to Loreville from In In theory, yes, so because we generate the navigation mesh only around actors and vehicles, right? And actually if we go farther away, we also destroy those areas at least not to to keep in memory blocks that are not needed. So, yes, you could go around the the planet with no problem. Theoretically. And in general theoretically, yes. If you survive them shooting at you, and yes. Well, we'll see we'll we'll how it we'll see how it plays out in testing and everything and there's there's a reason uh many games have leashes on Of course. on stuff. So, it's So, it's one of those things that it's
00:38:32
I think it's very likely that designers usually want them to stay more contained than than uh the actual technology preventing you to not do it, of course, because it's the experience you want to give to the player. A realistic character probably won't chase you for 80 mi. No, but because I've been told usually, yeah. Because I think they will be told to not do it and then I think eventually what we want to have is the is the ground vehicles, right? And then they might chase you with vehicles as well through roads. But but that's a wonderful illustration of the difference between a design decision and a tech capability and stuff. The the the the the the the feature, the tech that, you know, the team is building. Yes, it's technically possible and whatever, but then it will be a tech Sometimes it's the other way around. Sometimes my video is the other way where it's like, "No, you cannot do this by tech." And he said, "But I want
00:39:19
that." You know, so Sometimes Sometimes you have to be the bearers of bad news. Who? Uh, also recently in in ISC we've shown uh the work for AI reinforcements. basically going to one of these places that maybe doesn't have AI until you trigger something and then they come flying in uh on ships and stuff. Uh will AIs bring in large drop ships to unload both humans uh only humans or humans and potentially ground vehicles? In in theory, yes. At this moment we are testing only with the actors, with NPCs. also and also in general those Yes, those are big cargo ships that are part of the reinforcement behaviors, right? To to be
00:40:06
able to carry, I don't know, 20 NPCs, right? Uh I'm not even sure if capital ships could land on the planet. Yeah, anything could land on the planet now we detect that we made, but I think the real problem I think is only that um we, as you said, we're working on the navigation links, right? So, some ships have different methods to enter and exits. So, it depends on the ships that are the methods that we currently support that are the one with the back door right now. And the one, for example, that have internal elevators, we don't support them yet, you know, or something with the with the ladders, we don't support them yet. Um but, you know, the one with the back door, if it's a carrier ship, you know, then it's fine. If it's set up correctly, of course. But, that is just a
00:40:54
bare minimum, right? This might be more of a design decision, but as a follow-up, will the air reinforcements immediately begin destroying our parked ships, or will they engage players directly first? I think if it's not good if the player or the hostile is not actually in the pilot seat not try to shoot at them. not going to attack that ship. For now. Yeah, I think I think right now I think there's there's one thing I'd say we don't have a live a good tactic for them to decide how to destroy a ship. So, of course, you could shoot with a rifle, but it doesn't really damage much of a ship. Um and then yes, they would like it We also want to understand how it to expose based on the mission requirements,
00:41:41
right? So, for example, you the guys could could be called in not to destroy your ship, but to you know, destroy you or capture you, right? In theory, eventually. So, as I say, it depends on the type of information designers wants to base it in, and they would probably search for somebody. It it depends also what your reinforcement means, because I think you brought up a good point. You said it could be that you're right in a place where nobody is, and you trigger an alarm, right? That attracts some, you know, people that they need to go there and check what's going on. But, it could also be that there are already people that you engage, and some of them, you know, they manage to require assistance. So, they come and the people who come already have an understanding that something is happening because it's been communicated to them, right? So, you want to bring them with a different kind of alertness that is shown to the players. So, they arrive
00:42:29
already knowing that something went wrong with their with their friends and stuff like this. So, you know, we can support those type of things. It's a and then of course we've talked about, you know, AI with different levels of of intelligence and and personality stuff. Some you know, some you want AI who recognize I don't have weapons to take out a ship, so I'm not going to bother shooting at a ship. And then there's always got to be that one person who's, you know, new and not really experienced and he'll he'll take his little pistol and try to shoot down a cutlass, you know, try it. It's like Yeah, that that's part of building a dynamic and diverse universe. It's we you don't want every bot to be perfect. No, and I think it's also a matter that, you know, if it feels real, then it's
00:43:16
fine, right? You would expect if you shoot at the guy that is on the ground and he has a pistol, you know, he might shoot some rounds just to, you know, because that's what a normal reaction would be and then he tries to find a cover or while he you know, he looks around and he doesn't know where to find it. Still tries to shoot at you and then he goes and tries to, you know, find some some yeah, defensive area. That's where our our trait system will come in. We can have different traits for that be integrated with the combat behaviors and everything. So, if you've got a a gungho character or or a cowardly character, then there will be different traits and they will affect different roles on targets, how they select them and and other things like that. One other thing about the animations for people with guns is
00:44:06
what we get reviewed on a lot is trigger discipline and that's about as well how you actually hold the weapon and Steve Bender will chop you down if you don't understand proper good trigger discipline and it's awesome because we learn more about that that's actually the thing I believe that we had a one of the actors who was actually learning about using the weapons and because they were learning the performance actually came across as somebody who was learning so they weren't perfect and it was I believe it was exactly what was wanted we needed yeah exactly so yeah it's awesome the more real it is you know yeah the more real it feels we've definitely got some factions that have combat styles which are different
00:44:55
some are highly trained with their weapons and some are kind of like amateurs just grabbing their first weapons so that's definitely something that should come across with our animation style as well how do you know that's how I do Stuart with the weapons without jumping into spoilers there are multiple factions in Squadron 42 that you'll be squaring off against and they don't all have the same modus operandi let's put that way but even in the Star Citizen universe that the world is so large and vast right that that thing you will expect that there might be some settlements of people that are less trained or with some people that are less trained than others we're seeing a lot of the nine tails lately so this year with dynamic
00:45:44
with dynamic nav planetary nav mesh and with AI reinforcements we're going to see more and more hostile AI around the the planets and locations of the persistent universe this year than at any point in Star Citizen's history besides just fighting, uh will AI factions uh AI actors start using medical tools? Uh will AI run around with the med gun and start healing each other? Um it's definitely something that we've got planned for squad Well, I know if it is as being planned for Squadron 42 uh that we do want characters to be um healing other friendly characters. We also have designs in the works for um when you take down people with
00:46:33
non-lethal takedowns and you restrain them that they're both going to be um if that that character's being knocked out that they'll go and revive them and and uh release them from captivity. Um that's definitely something that's planned for Squadron 42, I know in in the design. So, that should also then come forth in into PUs as the rest of this stuff does. So, yeah, definitely. Uh how about allowing AI to loot dead players? You may want to I think everything is I know where everything is So, I think I think in general is is connected and we do the stuff that we want to do with the with the ammo and you know resource management as Jared
00:47:21
was also mentioning, right? For example, we do have already some work that we put in for investigating the bodies, right? And like in general is like when you will have to search This this all connects, right? So, it's like when you need to find ammunitions, for example, you finish your ammunition, what you would do is exactly the same as a bartender would do, right? That you would request the system to give you a usable that can provide you with a specific set of objects. In this case, ammunition for your pistol or your rifle, right? And that could be a body to loot. It could be a crate to loot, right? And that just means that you go to the object, you know, you play the animation that, you know, Stuart animation team makes. You know, you will look and you know, of course, what you are after for looting compared to that
00:48:10
you really need to go and find Oh, I don't know. I mean, you could do you could pretend those that you don't know and let the AI sometimes, maybe based on traits, you know, finding somebody that doesn't have anything as well. But in theory, you know, the AI might know where to to to go and what to loot. Um so, that these these type of things is an example where I think a lot of the work that we've done on the social side is not isolation. It's not It's exactly the same things that you need for combat. It's exactly the same thing for you that you need for pretty much anything. And, you know, it's And then it uses the routing as done as shown that is like the ammunition will be some usable inside a usable that is the body of, you know, that you need to loot, for example, or the crate. So, when you say pick up, it routes through the right object and then you play the right
00:48:57
animation because you carry this extra information with the action and then it's just a matter of setup, but the behavior is going to be the same. It's like give me something where I can get these bullets. Oh, is this usable? Okay, cool. Is this, you know, the It can either be a physical bullet that you find on the ground. Maybe you find a magazine on the ground. But it could be just that you know that something belongs to the loadout or the inventory of something. That actually starts to touch on one of the relatively new things that was was starting to look at, which is currently NPCs are all knowing, right? They know everything about everything. Like they know if the the wall panel that is on the the ship is in use or not. But what we want to start trying to
00:49:44
introduce is the illusion of them not knowing where something is. So, like Fran was just saying like with the the ammo, the NPC knows straight away whether that guy's got the right ammo for them or not, but we want them to go and check anyway. You know? And same kind of thing like um with something like a vending machine, if um if an NPC is going to go to a vending machine for something, it might be nice if another one comes up and then they go, "Oh, it's already in use. I'll just sit back and wait for a minute." And maybe that could even introduce like the the chance for for them to have some kind of a dynamic conversation as well. So, it's it's just those small steps that we're making towards trying to create this larger more complex AI behavior. So, they'll also they'll also never know if that vending machine is going to
00:50:34
Uh just torture your AI here. for for a little bit in the chat today, there was an interesting conversation about the nature of frustration going on between some members in the Twitch chat. And you know, if that's how that's how they want to spend their time, you know, so be it. But let's play on let's play on the nature of frustration and talk about performance issues. Um I anybody who's played Star Citizen for any amount of time knows that that that uh all AI in the game suffer from a number of performance issues. Um What of that can we fix today? What of that what what of that can can you know, can we work on in the short term? What of that requires waiting for things like
00:51:22
you know, entity streaming and server meshing and and and and and and and anything that improves overall server performance that's drastically out of your hands. I guess let me rephrase this I guess. What is within your control to help and and and work on to improve performance over time? And what do you just have to sit back being frustrated yourselves and wait uh for for other stuff to to come online. Let's talk about what you can do. So, um I I could start a bit on some ideas on what we are doing. We are working on per We have a lot of performance uh task that uh we try to tackle and to to improve the the frame rate for the um for the dedicated server because on the client is not a big problem. Usually you
00:52:10
stream only what is around the player and is you render only that. So, it's not that costly. On the dedicated server because we don't have yet the the server meshing. If you have 50 players and each of those 50 players are in heavy areas like Lorville, Area 18, New Babbage with a lot of uh NPCs, right? It happens that the dedicated server is going to have seven five frame rates per second and this is going to impact the how fast and um you are going to see the decision from the NPC. So, yeah. This is one of the the biggest uh problem that we have. And what we do is working on performance task where we can. Plus, we try to
00:52:58
discuss with directors, designers to try to limit the number of entities that we put in the game. So, yeah. These are the the two things that I could think at this moment. I think that definitely uh sorry, goes to you. Um one thing and definitely correct me if I'm wrong, friend, but um we when it comes to the performance of the characters that are inside the usables, we do have this sort of thing that we can fall back on and I believe it's called usable idle. And that's the technical term anyway, but basically if the animation is um having trouble loading uh or there's too many NPCs and then there's just one that's just asking the server, "Hey, what's my next animation?" And it's just taking a bit of time to There is an idle animation which I will
00:53:49
put together, which is in context to that particular usable. Um and then if even if it's not loaded the animation, at least it's doing some sort of a performance that's in context. Yeah, yeah, yeah, basically you said that it's not about loading, but it's deciding what to do next, I think, right? That's what I think is what probably you meant. And I think yes, absolutely. I think it's a lot of the stuff is being smart, especially in the social environment where you know that the player is not going around and killing everybody, right? Because they need to react very fast and so on. But I think in the social environment, yes, we need to absolutely be smart on the way we cover up the the time it takes to find something when there are like 1,000, you know, 500 characters acting at the same time. That's just a lot for one server, right? This is where I think it's even pretty good for what we have
00:54:39
right now. And of course, I think still the experience, you know, it's not what we would like that the player would have. Um there is a bit like more stuff that we are discussing right now. I mean, first Mars is probably up three, right? When he says like I mean, we do a lot of performance capture right now every time and we see where is the biggest bang for the buck as well because sometimes I think there are I mean, AI itself is definitely not that expensive, but it's everything that comes with the character, right? That is animation, updating animation, the rendering, uh you know, all the calculation for I don't know if the pathing of a ship with all the items and stuff like this. So, it's AI is at the end of the day is the end of the you know, at the end of the all the chains of all the features. So, to have good AI, you need to have everything else that works. I think one thing is
00:55:26
definitely like, you know, the desynchronization as well that we are looking at so that we try to fix bugs sometimes. I know people feel like, "Oh, this is what happens on the game." But I think, you know, it it might be different. Some of these bugs are related to the fact that you see on the client something different than what happens on the server. And like recently, we have fixed a lot of those bugs where for example, there was a problem with the lower body turning. And the fact that the lower body, you know, was not turning on the server, you would see a character fighting that he was not looking at you. In the server, he was trying, you know, but it was not turning. And that information was not arriving to the client. So, you feel like, "Oh, the guy is not responding to me because he's not shooting." He's not shooting because he's not pointing you the gun because he can't. And on the way in the server, he can't the client he doesn't even look he's even trying, right? So, I think you have sometime with this discrepancy on
00:56:14
experience that we are These are the bugs that we absolutely are looking at right now. Uh we don't really need to have any change uh from server mesh or persistence or any of other type of dependency. I mean, it it's time, as you said. Sometimes, you need to just pick with the balance. Um but there are other things that I think we want to do. For example, in last year, uh last at the end of the last year, some things that we changed is the way we give priority to some of the calculation. So, we give calculation to the combat more than the social and to some things at the social a bit above the combat. And uh you know, I was finishing to experiment with some other ways of give priority with a lot of different LODs, you know, um and uh these are stuff that we we have. We have
00:57:01
already a tools version LODs on behaviors and movement and stuff like this. But we don't have yet like for example, what they want to do is a nice way of balancing importance of the request based on the LOD you are. And if you switch LOD, right? If the player moves around and fights with uh you know, let's say with Andrew, right? And and then he's super far. Maybe he was close to me, but I left him, right? So, if now it takes a little bit of more time for him to find a new one cover spot, I prefer that Andrew's cover spot is quicker to give to him. And we try to be very careful as well. We have a really limited set that of ray casts we do every frame, right? Or a very limited set of amount of of you know, queries that we do. And some of the low system that this query system use, we are trying to optimize as well. Like the usable queries we we we are right now working on making them faster.
00:57:48
If that becomes faster, you can run quicker queries, right? So So so there's a couple of things you can still do. And will the more global things like like server meshing or anything, will that help AI at all? Absolutely. Because I think like I think the best thing that we reduce the workload on the servers, right? And I think as a lot of people report, you know, when we read those their part of you know, of Reddit or everywhere is like you can feel that there is a very good experience from people also when the servers are are fresh, right? And there are not a lot of systems or a lot of locations enable. So I think this is why we are really pushing through this technology that is the server mesh because we know by a fact that if you run you know, more servers with less you know, workload, the experience is closer to what regular
00:58:35
games, you know, do right now. And just to talk about briefly about our kind of frustrations as well as like we we could make a thing and we'll test it in single player and it'll work great. We could test it again in a local server on our machine and it'll still work great. But then you send that out to the PU and we could do later you get a bug back saying it doesn't work in the PU. I was like, but it works. Um all right, so we are actually out of time. Uh but we're let's do a let's do a lightning round here. Let's see if let's see if we can squeeze in uh a couple more questions here. Lightning round, short answers. Fran. I need you to try man. I need you to try. I'll shut up then. Uh are AI able to use grenades currently?
00:59:23
They are. I think my self been working on that long some time ago, I think we want to refresh a little bit that that functionality. Um will there be frustrated animations for NPCs who miss public transport? Yes. Yes. That That will be fun. Uh anything you can tell us about animal NPCs? Fauna. Every Every animator wants to work on Just a matter of getting them prioritized, right? We We've definitely got some uh work coming up shortly for some uh animal NPCs on Squadron anyway. So, hopefully that should feed back into the PU. Yeah, we have some in general some stuff also work in progress for and you know, fauna but Okay. Uh uh the AI turret blades for for spaceships just to have not like NPCs in
01:00:12
them, but computers controlling turrets. So, that that is already working basically. It's what we call the AI modules. Uh I think it's not exposed. I think this is a gameplay uh you know, question to how this is going to be exposed to be you know, to be able to buy them and apply them in your ship. We do have that It's the tech working when players are facing the the the the pirate Idris and the pirate Javelin and stuff like that. That's how those are Yeah, for example, the unmanned turret what they call unmanned turret are actually turrets with an AI module attached to it. So, it's it's it's it's just has to be exposed to the player which is a whole another series of design and UI and all kinds of stuff, but out of out of your hands. will the AI have real feelings?
01:01:02
You got to click answer. We'll do the Turing test by 2040. I think I think there will be like um like a way in which the AI will express their mood, you know, as we were saying before like frustration is one and how this this mood will be consistent, right? There is this this concept. Um and you know, of course, they will be hurt if you insult them, right? uh for your area of expertise, what's the biggest challenge right now? What's your al wapo? Set up and consistency in setup is quite challenging. Especially with a big big studio like us, making sure everybody's doing things the same kind of way. Cuz when you go back to debug something
01:01:51
and it's set up entirely different to how it should be is yeah, well, it's another frustration. about Stewart, aren't you? I can tell. No, Stewart's actually really good. He's yeah, he's he's great. Anyone else? Biggest challenge in your job? Getting mocap time, but that's going to be fixed. For me is the is the inter like intertwined systems, right? That they have for example, just the usage So the hours looking at with Marius, right? Was this transit system, you know, and then you know, you need to have a full understanding about how system works so that then they I can also use it in the most consistent way, right? So sometimes you need to have some either dedicated time from another team and you know, making sure that those things are synced
01:02:37
or make your you know, the best of your guesses, you know, on how something works so you can collect the right questions to use a you know, the other team's time you know, in the most efficient way as possible. And then the fact that sometimes some of the new feature this might take a list that they usually have an impact, right? So it's like you you do you need to do stuff in a way which encapsulate the risky part that might be affected to the minimum, right? So that as soon as something happens then that the all the it's all in that side. For example, the load you know, the load outs and inventory, we have been talking a lot about the the the lockers, right? We did an implementation that way which is like you they are going to go in the locker and they can change now as soon as that the inventory right is there and the functionality for the locker ah we
01:03:24
can just replace the little part. Andrew, where is Uh Andrew. Oh, I was going to say lots of planning meetings but I think yeah what I said before complexity of sometimes complexity of systems just having to think through everything so you can get stuff right as possible the first time is that's probably a big challenge for me. And Marius. Uh challenging but in a good way is working on and trying to find solution for performance. It's something that gives you the the best for for that situation especially with your with limited resources. Well, that's it. You did it. You
01:04:14
survived You survived the Hour Long Show. Thank you so much everybody. For those watching at home remember Invictus starts today. I think it's probably starting like right now if I'm not mistaken. Be sure to check out Crusader all the stuff that Invictus this year has to offer. It'll be changing up every day. Keep an eye out for Drake Defense Con this year and some of the folks who are around in the EU hours this morning may have gotten a slight unintended preview of something new and very exciting coming at the end of the whole thing. So so stay tuned for that. Keep an eye out on the robertsspaceindustries.com website for for all the details including a bunch of different contests and stuff that are
01:05:03
going on your chance to win all variety of stuff which I say cuz I have no idea what you win. That's that's a whole another department and I forgot to ask before I did this because I'm a professional. Uh come on back to us uh next week on ISC. Uh we'll be looking at the new cutest vehicle in the first citizen universe, the Drake Mule, uh and uh the arrival of of of one of the most exciting fighters uh in recent memory, the RSI Scorpius. Uh and we'll be stepping foot inside our new uh UK studio uh for the very first time. Uh so yeah, so check that out and then we'll be right back here next Friday for an Invictus Launch Week all ships Q&A. We'll have John Crew and members of the vehicle team uh to answer questions
01:05:50
about all the spaceships and vehicles and and stuff that are that are fit to print. Uh so yeah. That was uh Stewart. That was Andrew, Dan, Francesco, Marius. Thank you so much for taking the time again to be out of the out of your end of the week. Um and uh for those in the chat who spent like a good 10 minutes just debating the nature of frustration. I don't I I I I I don't know what to say. I don't I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I Not any place that's not going to get me fired. So I'll just say good night. Take it easy everybody. Bye. Bye everybody. There was literally like four people
01:06:38
just going back and forth about the nature of frustration in chat for like 15 minutes. I'm just like, "Oh my gosh."

Transcript from YouTube’s automatic captions — punctuation and Star Citizen jargon are approximate. The video itself is hosted by CIG on the official channel.