Due to heavy work commitments my Armanda set has only been able to act as desk ornaments (which they do very well!), so sorry if this is a silly question!

There is an OK X-Wing scene where I live, but it’s far from great. Consequently, I’ve used the X-Wing AI a fair bit, and have found it very helpful for practising with (http://s-fear.com/xwing/). I’ve been asking around, and while there’s a bit of interest, Armada players are going to be very few and far between for the foreseeable future.

For those who actually know about creating something like the X-Wing AI (of which I can’t count myself!), do you think an Armada version would be possible?

Cheers

A tad off subject here but I played a scenario mission with myself where the Imperial Victory Class Star Destroyers simply flew in a fairly straight line with random fluctuating built in. It was interesting but not ideal.

If you can teach an AI to make intelligent use of the maneuver tool, you're a genius.

AI= Ain't Intelligent

No AI needed fly at speed 1 with repair command spammed!

I'd be happy with a play-by-email version.

Need a gui to set things up for firing and maneuvering, have the rules encoded, die and card generator, and an output that can be emailed to someone. Their gui would then have to validate the file it go against its last known turn with the list of actions taken by the other player. The gui could even output exact coordinates, so players could mark off the game on their own personal playmats.

So is it only the maneuver tool that would stop an AI being creatable?

Like the idea of an email exchange system, but not sure what a gui is?! :-)

Vassal would work when you're able to sit down at a computer to play, and would have an interface (GUI) as well as all the tools needed to play (aside from full card text, as that tends to be omitted to avoid copyright issues). You're also playing against another human to boot. Check out ransburger's thread here:

https://community.fantasyflightgames.com/index.php?/topic/141256-vassal-module-timeline/

One video game programmer's perspective:

The maneuver tool itself is not an impediment to creating an AI. The maneuver tool creates a finite set of locations, each one of which can be evaluated for suitability; this is how a lot of game AIs work. The number of possible locations is small enough that any modern smartphone can compute all the possibilities and evaluate them very quickly, assuming the AI has all of the information available.

Getting the AI that information is the problem. You're not going to want to input it manually every turn, that would be too tedious. You could just input the starting positions and then the maneuvers each turn, but the computer model is going to diverge from reality due to humans not placing things precisely, and that will probably also get rather tedious.

I was thinking about a method in which you put a QR code or something similar on each ship. Then, when it is the AI's turn, you take a picture of the board so that the AI has all the positioning information instantly. I have never worked with cameras or imaging systems like that though, so I don't know for sure if that would work. The image from the camera might be too distorted to get useful distance information from.

Making a play-by-email engine would effectively be making a digital version of the game, which is really not very difficult as video games go. FFG has sort of done that with Battlelore, but I don't know if the Disney license would allow such a thing for Star Wars. I suspect the dice roller is the only Star Wars app for a reason.

Thanks Brian_Black :-) I tried to give the X-Wing version a half-hearted go, but I think with the lack of Armada players here I'll definitely I'll try again.

Instead of exact positions, maybe the AI could operate off of ships being located in hull zones and ranges, along with speed and orientation? Might reduce the "smarts" of the AI, but it alleviates the data entry burden of the user.

The big problem is squadrons, as there is a significant strategic element in their positioning as fighter screens and the choice of targets to go after. I.e. you have four TIE fighter squadrons and there are two seperate X-Wing squadrons, do you send all four TIEs to the closest or send two to each? thats the sort of decisions players have to make in Armada and a "simple" AI going "do this manuver" is not going to be able to replicate that.

It's possible for the AI to operate with less information, yes. Whether it would be "enough", I'm not sure. It might result in the AI crashing a lot. It is also possible that distortion in the picture could actually be advantageous in this case - it could let the AI make more realistic mistakes and act more human.

I guess it depends on what you expect - are you just looking for something that isn't completely dumb to test some ideas against, or are you looking for a challenging game experience?

If this is an actual project of some sort... you have to start somewhere. Not much would get done under the development instructions of "Deep Blue or nothing!" And yes, it's much harder to code moderate/dumb intelligence than a smart one, assuming the AI model has access to all information about the situation.

GUI = graphical user interface. What you see on the screen.

A play by email would be super casual. Take at least a week to resolve a game. Maybe more because of the casual nature. Have to take turns setting up obstacles. That's a couple of days unless online at the same time. You set an obstacle, measure coordinates, email it. Then get their data and set it up. Then another day or two for the ships. Then each round might be a couple of days depending on the ships. The game could take weeks, unless two dedicated players were working at it each day But, for some people, it might be the only option.

But with a play by email system, a graphics user interface could be made to play the game on a screen. Problem is, the interface would have to augment the miniatures FFG wants you to buy, not replace it.

I witnessed someone playing chess by fax. It was a gentleman's agreement that each would only take 30 minutes after reading the fax, to write their movement and fax back. The chess match was in its 40th day. Might be a day or two between moves. The "chess table decoration" was really a slowly ongoing game.