Two Seater Challenge

[Peyoteros]

So, this is it? That's how much everyone wants TWO SEATERS? No more ideas, suggestions? :huh:

[VTJS17_Fire]

I have some ideas. ;)

 

 

+ If no mate is online to fly with me, I should choose, where I can sit in the plane. Maybe, I want to fly as RIO/WSO and the AI is my pilot.

 

+ But, if one of my mates comes online, he should take place in the free seat and can replace the "AI mate".

 

+ If I'm the pilot, I should give attack and other commands (target sorting) to my AI RIO/WSO. The human WSO of my wingman (element lead) should also give commands to my WSO.

 

+ It should give an ingame voice tool ... an integrated TARS. I've should talk intern with my WSO/ pilot, without blocking the radios (real feature?).

 

+ The systems of the pilot and WSO should be synchronised/ work together as one system. If my WSO sets the radar, the settings on my screen should be the same. Same for the TGP.

 

 

kind regards,

Fire

[Irafos]

In my mind, you really only have two options. You either release a game that contains two versions of the aircraft, i.e F/A-18 C/D or you do what ArmA did. ArmA II allowed the pilot of any two seater airframe to take over the firing controls of the gunner in the event of a gunner absence, which wasn't as effective as having a a competent WSO with you, but at least you were able to play. The second and most glaring issue with the whole thing is the Top Gun issue as mentioned above (no one wants to sit in the backseat, everyone wants to fly). Well, you can circumvent this by NOT going for a two seater such as the F-14. The F-14 contained a pilot and a RIO (Radar Intercept Officer) who didn't really manage any weapons and was just another eye and hand for the pilot. In the F/A-18 and the AH-64, the roles are a little more defined and diverse. In the F/A-18, the pilot has controls over flight and navigation and the gun in most cases, however, the WSO (Weapons Systems Officer) has control over all the weapons and in some cases, countermeasures. Thus, this allows for a more integrated play experience, rather than having one guy watch the other guy do all the cool stuff. Another marketing option is to do what Portal 2 did. In Portal 2, the Co-Op multiplayer was much different and had different content than the single player. A good idea is to have the inability to fly backseat in SP (only fly front seat or have a one seater variant) and add the ability for a back seater in MP, thus making the backseat accessible only in MP and giving players incentive to jump in the cockpit with another bud and learn the backseat. Any thoughts?

[Bushmanni]

Some thoughts on AI as a co-pilot/gunner. Flying with AI is to work as a team with AI. Teamwork is about complementing each other with your actions so as for the team to be more than the sum of the individuals. This gets done through coordination with your buddy and that is done through predetermined actions, planning and communication.

 

Predetermined actions would probably have to be created by the dev team as I can't figure out a easy enough system for user to build them for himself. These actions would have to be described very thoroughly in the manual and how to use them and how not to use them. Basically they would be some simple operating modes (fly for me a while in current direction and altitude, set weapons for ground attack, scan for targets, etc.), tactics (orbit target beyond small arms range, get in hide and start doing bob-ups, etc.), ways to prioritize targets or own actions (kill AD first, kill tanks first, never ever get in range of enemy, engage and close to the enemy, etc.) Player could order the AI to change between these actions through communications UI.

 

Communication is bi-directional. Player needs to have means to understand what the AI is doing, what he's going to do and what's his problem when he's not doing what he's supposed to do so the player can fix the situation and enable the AI to do his job. This is a pretty complex thing to accomplish and requires some psychological insight to make it work. There have been some research and implementations of this kind of idea in US Army's simulators and the catch phrase used for this kind of stuff was "explainable AI".

 

Communication itself would ideally be done by voice recognition and speech synthesis. Using keyboard to issue orders might be doable but would need to be done lot better than it's currently done. SWAT 3 and 4 had pretty well thought out comms. interface but then again that might not be doable with an attack helicopter sim. The basic idea of SWAT comms. interface is to put the often used and "life and death"-important orders under one keypress. The ones that don't need to be issued ASAP to be meaningful were hidden behind multiple levels of menus. The other brilliant feature was that the available choices will change depending where the player looks. If you look at door you get stack, cover, breach, etc. orders, if you look a civilian you get detain order, etc. Context sensitivity might not be very useful in flight sim as the important objects tend to be rather small and hard to point at accurately.

 

Players have the TIR as excellent pointing device that could be utilized in AI comms. interface. This is especially realistic for AH-64 where the pilots can see where the other is looking through their helmet sight. Instead of having complex orders selected through menu you could have few buttons assigned orders like fly there, attack that, orbit that and look there.

 

Voice recognition is already used so there's no big steps needed here. Basic problem is that when you have lots of different commands, it's hard to remember the exact phrase that triggers it. This probably can be sorted out only by having a phrase list always available when playing or memorizing all of them. More ideal solution might be a "dictation to text" software and parser that could make sense of the words the user has blurted out and take into account the context of the situation. But this is the future, not the present.

 

Because the AI needs to have lots of things to say in order to be explainable, using pre-recorded voice messages is impractical. We have already seen this in JTAC as it's not getting new features added to it as it would require new messages recorded for it. There's already available quite good speech synthesizers, also available as API that could be used instead. This way adding new phrases is just adding new text strings outputted by the AI for the synthesizer. Check this out. https://acapela-box.com/AcaBox/index.php

 

Planning is a tougher nut to crack as you would need an AI that's capable of understanding plans and intentions instead of having simple action sequences and conditional branhes in it's logic. The basic idea for humans in planning is to build common picture of the situation and guidelines, principles, intentions and procedures to overcome the problems the situation presents to accomplish the objective. It's like this as a consequance of how the brain works but the AI logic doesn't work like a brain and hence is incompatible with abstractions of human though unless made to simulate it. I can't see this happening for quite a while in general scale while it might be already possible with some experimental computer systems.

 

In short, build an AI that can communicate back to player effectively and provide information the player needs, preferably without asking and at least when asked about, and a communication UI that's fast and easy to use, utilising the best methods utilized in current games.

[EAF51_Voss]

Greetings to you all,

I think that would be a wonderful thing by third parties, could get a dual-control trainer jet. I think this is the best gift you could do the whole international community and individual Virtual Squadron.

 

Nik

[cichlidfan]

Greetings to you all,

I think that would be a wonderful thing by third parties, could get a dual-control trainer jet. I think this is the best gift you could do the whole international community and individual Virtual Squadron.

 

Nik

 

I know the T-38 isn't exciting but it would be interesting from a realism standpoint, especially for someone new to jets or flight sims period.

[Peyoteros]

Could be L-39...for a start. Nothing secret about this one anymore, hence there should be plenty of info on it. From my point of view, that would be one of the best candidates for the first two seater attempt. Personally, I'm very big fan of Yak-130... :P

Edited May 7, 2012 by Peyoteros

[EAF51_Voss]

Could be L-39...for a start. Nothing secret about this one anymore, hence there should be plenty of info on it. From my point of view, that would be one of the best candidates for the first two seater attempt. Personally, I'm very big fan of Yak-130... :P

 

Hello and thanks for your answers, well I would say that would be fine as well as the Aermacchi MB-346, but the discussion is the feasibility of having a model airplane in all its functions, two-seat multiplayer and the ability to have two pilots (instructor and student) with the ability to replace the controls, any time you deem appropriate.

 

I think this is the essence for us simmers. :thumbup:

 

Nik

[Exorcet]

I don't put much priority on 2 seaters, but I can see why they would be fun. From a sales perspective, it would probably be best to release a two seater as an alternate version of a single seater. In other words if the next DCS is F-15, we would get a C and D or C and E, or something along those lines. This way, you don't need to worry if the 2 seat crowd is too small to support a game on their own as both single seaters and two seaters will buy it.

 

As far as implementing two seaters, both seats should be available at all times. I don't see the point of making one of the seats online only Just fill the missing seat with AI and possibly allow switching for the player. In online missions, the host could specify if two seaters allow switching and AI

[Pikey]

OK so lets consider 2 seats and lets pick a non helo only because we have Black Shark.

 

Let's first start with one player. What do we want this AI to do when we switch to the RIO? Auto engage Autopilot, go slack, create a super AI or allow you to control the plane in the same way without the same vision or instrumentation?

 

Which one of these fits the simulation? If any? Or is there another option?

 

Now lets take two players? Can we find a "rear seat" role/job that is engaging in the same way as piloting? (I can imagine this in some chopper sims, we can come to that)

If so, can we list some airframes to shortlist and get some expert knowledge going on about them (because I haven't the foggiest)

 

OK so it seems to me Helicopters are somewhat uniquely interesting in the 2 seat role. Especially as a gunner. A question...are most of these weapon officers able to use dual controls? As a single player the same questions above about AI.

 

At the risk of sounding negative, people talk about AI as if it is something easy to implement, but don't forget how hard a Chess Supercomputer can think and the best humans can still play on an even keel etc. I mean how do you explain to the AI you are creeping up on the ridgeline and wanting to pop up? Or how can you say quickly to take the northern Truck and work right, especially without slowing the entire process at the risk of being shot at? What kind of experience will you get when the AI does something you didn't approve of? And what use is an AI if it cannot take independent actions without approval?

 

And if you could have an AI like that, it's not a simulation, you are just building a super-AI which isn't modelled on the airframe.

 

It still seems to me the choice is between a 'pure' 2 part multiplayer sim for a small subset of an already niche genre like VFR members, that doesn't give equal experiences for each player or rely on what would surely be something artificial right from the design start point.

 

I've not mentioned my personal feelings (until now) because they have little value. But I will to help illustrate my argument direction:

 

1. I'm not in a VFR, if I do play any MP it will be on a casual basis much like the old IL-2 days but thats unlikely outside of a scenario requiring much in the way of teamwork and coordination - ie furball classics might attract me or playing OPFOR. I haven't done that for years anyway and frankly the level of coordination is intense and the starting skill point quite off putting.

 

2. I want to pull the trigger. It's my trigger. Likewise, I want to steer. And I want to select my own targets. I dont want to be staring at a radar screen for 30 mins.

 

3. I don't want a super AI. Because i'm better right up until it comes to enemy detection. If the plane doesn't have it, I don't want it either.

 

Basically if a dual seater was implemented faithfully, it's unlikely it would be something i'd buy.

 

But this is not to say there aren't developers who work closely with others to test this and there couldn't be a 2 -seater for those that already have their Maverick or Goose lined up online and available.

 

That all sounds negative, but in reality i'm putting the issues out there for solving because that's what the folks who want this need.

[ralfidude]

Would a two seater F15E be possible, as in declassified?

[Peyoteros]

I don't know guys, but AI copilot for MP, makes NO SENSE at all... It's MP we're talking about. Why would anyone ever...?

 

AI copilot in MP defeats the very purpose of two seater plane IMHO. One would need AI assistance in controlling two seater in SP mode only.

 

As for the MP, well, if there is no partner to fly with ATM, fly with random player who wants to fly with you. Don't feel like playing with random Homo Sapiens, but feel the urge to get involved in "two seater action" right now (OMG, I can't wait any longer... come ON! :furious: ), play SP while your mates are back from work/school/pub . If you wanna fly two seater with AI only, then, there is a BIG chance that you are SP kind of guy (girl), then, there is no business for you to be in MP part of the game in the first place, stick to the endless SP fun.

 

There should be some sort of script which would prohibit two seater from spawning in the MP game, if copilot seat is empty.

 

We have plenty of single seat airframes for the taste of anyone. The very purpose of two seater is COOP. I believe, we don't need to worry about AI copilot for MP at all.

 

What do you think?

Edited May 8, 2012 by Peyoteros

[Daze]

Could be L-39...for a start. Nothing secret about this one anymore, hence there should be plenty of info on it. From my point of view, that would be one of the best candidates for the first two seater attempt. Personally, I'm very big fan of Yak-130... :P

 

Start emailing Lotus Sims asking him to contact ED to try and port the L39 into the DCS engine!

[Wichid]

Start emailing Lotus Sims asking him to contact ED to try and port the L39 into the DCS engine!

 

I think you'll find they are creating planes for Microsoft Flight now... oh wait.

 

zing...

[Peyoteros]

Come on people, it's two seater we're talking about! Where is your enthusiasm? :huh:

[Chazz_BMF]

What about the option to steer the AI pilot using realtime orders like in other shooters like arma or ghost recon. Simply click on a point in space or on a map, then via a menu type thingy give the pilot a ROE. The AI should offcourse have its own evasive action to threats like AAA. You wouldnt be in control of the plane but at least tell the pilot how to react in the terrain fly low, avoid radar, seek and destroy etc etc. the same way visa versa when flying as a pilot to control the gunner.

[Peyoteros]

Would be nice to hear the word or two on this matter from the DEV's or testers, or high ranking officials from ED or TFC ... What they think would be a problem in creating the "Two Seater"? I really think it is time to push DCS to whole new MP level. Meanwhile lets talk about it.

[Bushmanni]

I don't see any problems (at least technical ones) with coop two-seater as it's already done successfully. The only possible technical challenge not already done might be syncing complex modern avionics of the both players together as that might need lots of bandwith or then not, I don't know. But otherwise MP two-seaters or multi-seaters are no problem but maybe conceptually, ie. how to provide good experience to players.

 

One solution to problematic multi-seaters like UH-60's doors gunners is to provide the lesser exprerience for free as RoF does with rear gunners. You could always get your BF3 or Arma buddies to be door gunner for you if it was free.

 

You can't operate Apache or similar aircraft with one seat empty so you will need AI co-pilot for SP but of course the MP coop is why everyone (I'd think) wants to have two seater. If you don't need AI for SP (like in F-18D) there's really no problems to create a two seater. But people who want Apache or other attack helo in addition to Shark the SP AI co-pilot is a challenge that needs to be solved. AI co-pilot is also good to have for SP training even if you would mostly fly online with a buddy.

 

Current problem with AI is not so much the CPU power required for it but building the logic within reasonable time frame and updating it without breaking it. You would either need simpler AI system or better tools to manage more complex rules. Currently there are not better ideas so the better tools is the obvious answer. Industrial logic controllers are nowadays programmed with graphical UI. It's bit like drawing flow charts and connecting them with lines and picking rules or actions from listboxes and adjusting their parameters. And if necessary you can add scripts if the ready made tools won't cut it. This results in a more comprehensible logic which is easier to debug and update. Creating a logic editor like this is a big effort but in the long run it would most likely pay off. It could also be something that an end-user might be able to learn to use to create his own AI behaviors. A good AI co-pilot needs to have quite complex logic rules to not be a liability.

 

Pikey: How I understood you was that you are not happy with a bad or good AI co-pilot and I couldn't comprehend your rationale for it. Do you mean super AI as a cheater or as something actually capable?

[nemises]

Current AI of friendly AFM units (say A10c) is pretty good...

I believe he uses all sensors as a player would (probably slightly simplified, but within the same parameters at least)..

I don't think that having AI take one of the seats is a technical challenge for ED...and the command structure could / would flow just like the F2 radio menu...

 

So...at a guess, I'd rekon the only thing stopping ED doing a 2 seater, is the fact that they have other stuff to do first! (or do they?!...who knows what the next unit will be!)

[213]

they're doing edge/nevada, dcs: fixed wing jet, and multicore

[slowhand]

:beer:I think that a 2 seater jet or heilo would be a great idea most of us have a dark side that only your #2 can bring out..I could see a few good drinking games come from it...:drunk::chair:

[Exorcet]

OK so lets consider 2 seats and lets pick a non helo only because we have Black Shark.

 

Let's first start with one player. What do we want this AI to do when we switch to the RIO? Auto engage Autopilot, go slack, create a super AI or allow you to control the plane in the same way without the same vision or instrumentation?

I assume it would just be regular old AI, the same one that flies AI jets.

 

I mean how do you explain to the AI you are creeping up on the ridgeline and wanting to pop up? Or how can you say quickly to take the northern Truck and work right, especially without slowing the entire process at the risk of being shot at? What kind of experience will you get when the AI does something you didn't approve of? And what use is an AI if it cannot take independent actions without approval?

 

And if you could have an AI like that, it's not a simulation, you are just building a super-AI which isn't modelled on the airframe.

Same way you give wingmen orders. AI is AI, we already deal with them in the sim, so it's probably not going to be a huge jump.

 

 

I don't know guys, but AI copilot for MP, makes NO SENSE at all... It's MP we're talking about. Why would anyone ever...?

On the off chance it's helpful. I mean, why not have it? One of the most frustrating things in games is when an option is greyed out for no reason. If there is AI co-pilot in single player, it should be in multiplayer. The only reason not to have it would be the time it takes to code and work out bugs, IMO. In the MP situation, it just serves as a filler anyway. What if you want to fly XXX, but no one wants to co-pilot it? Without AI option, you can't join the MP mission, and there is more reason to MP than to just have another human in the cockpit. Plus, it's probably less boring flying with AI than it is sitting around waiting for someone to show up and fill the seat.

 

 

 

What about the option to steer the AI pilot using realtime orders like in other shooters like arma or ghost recon. Simply click on a point in space or on a map, then via a menu type thingy give the pilot a ROE. The AI should offcourse have its own evasive action to threats like AAA. You wouldnt be in control of the plane but at least tell the pilot how to react in the terrain fly low, avoid radar, seek and destroy etc etc. the same way visa versa when flying as a pilot to control the gunner.

 

This sounds like a good idea. I was just thinking expand the wingman commands we have. Instead of:

 

Flight

Wingman 1

Wingman 2...

 

we would have:

 

Front/Back seat

Flight

Wingman1...

 

The first option would have it's own special commands in order to allow realistic and coordinate communication between player and AI.

[Peyoteros]

I rather doubt it, that you'll have to wait for copilot for a long time. The way I see it, it's much more likely not to have two seaters available... I'm not a coder, so I wouldn't know how difficult would be to make a medium intelligent AI for the MP. My mission creating experience shows that AI are very likely to go funny as soon as they hit MP... Most of the scripts just refuses to work the same way they do in SP. Anyhow, would be good to have someone with coding background to say a few words on this matter.

 

On the off chance it's helpful. I mean, why not have it? One of the most frustrating things in games is when an option is greyed out for no reason. If there is AI co-pilot in single player, it should be in multiplayer. The only reason not to have it would be the time it takes to code and work out bugs, IMO. In the MP situation, it just serves as a filler anyway. What if you want to fly XXX, but no one wants to co-pilot it? Without AI option, you can't join the MP mission, and there is more reason to MP than to just have another human in the cockpit. Plus, it's probably less boring flying with AI than it is sitting around waiting for someone to show up and fill the seat.

Edited May 9, 2012 by Peyoteros

[Bushmanni]

The current AI and comms UI won't cut it for AI co-pilot as you will need to work effectively as a team with it. If your wingman messes around unintelligently it's not that big problem as you can still operate effectively yourself. But if you have a stupid AI who doesn't know what he's supposed to do and cannot communicate effectively what he is doing and seeing he will mess your performance also as he's operating critical systems you have no control over yourself. And the current communications menu isn't very fast to use and will require removing your eyes from action which are both bad for combat effectiveness. This would be especially bad for communicating with co-pilot as you will need to use it very often.

[Nate--IRL--]

they're doing edge/nevada, dcs: fixed wing jet, and pink elephants

 

Why does everybody forget about Combined Arms?

 

Nate