Jump to content

FAR Fighter Challenge: BD Armory AI

Wanderfound
 Share

Recommended Posts

Phearlock

Phearlock

Posted
Posted
Just had a test with an updated FAR; as far as I can tell, the MiK is flying the same as before. Did you have Dynamic Deflection installed on your test?

Yup, though the Pidgey handles differently on my install compared to yours as well (everything pulls significantly more Alpha when turning in my install, both the pidgey and the Mik-4, hence the MiK-4 spinning out of control at low speed and the F3C handling differently). So I think there's something weird going on. I'll double check I have everything updated and then see how the AI flies both again, if it's still different, I can record a vid of how everything handles on my end.

You can easily check if dynamic deflection was installed by simply seeing if the values have been changed (For example the pidgeys stabilators should go from 55% at 0 dynamic pressure to 100% at 40 then back to 50 at 160.)

Link to comment
Share on other sites
Wanderfound

Wanderfound

Posted
  • Author
Posted
Yup, though the Pidgey handles differently on my install compared to yours as well (everything pulls significantly more Alpha when turning in my install, both the pidgey and the Mik-4, hence the MiK-4 spinning out of control at low speed and the F3C handling differently). So I think there's something weird going on. I'll double check I have everything updated and then see how the AI flies both again, if it's still different, I can record a vid of how everything handles on my end.

In the interests of eliminating variables, I've created a new KSP install and stripped the modlist down to the minimum needed to video the fights:

EUNRwoc.jpg

The mods are all freshly downloaded just now, and all general settings are left at default. If folks use something similar while they're flight-testing, hopefully we should be able to get consistent performance.

- - - Updated - - -

Just tested the new install, and yes, the MiK-4 is overcontrolling. Looks like the weirdness was on my end; no idea what the issue was (maybe my overall Dynamic Deflection settings had been bumped off default?).

Anyway, hopefully sorted now. Assuming that things are finally as they should be: did you want me to run the F-3C against whatever I come up with after I get the MiK-4 tuned back into shape, or would you like to revise it first? The one place where the F-3C is at a definite disadvantage is in fuel capacity; those twin engines limit you to not much more than 10 minutes of airtime.

Link to comment
Share on other sites
Phearlock

Phearlock

Posted
Posted (edited)

None of the tests I did ran long enough for the Pidgey to run out of fuel even if it was only fueled to 50%. If I run into fuel becoming an issue I'll bump the tanks up to 75% and maaaaybe 100%, but I don't see much reason to do so yet. Now that your version is corrected the Pidgey shoooould fly as it was designed to.

Assuming your aerodynamics are correct the AI should pull to about 15(when fully loaded) to 35 (low fuel & no missiles) AoA normally in the Pidgey, with a tendency to stall it a tiny bit in the 240-180 m/s range when fuel levels have been expended slightly. (It's designed to be slightly more maneuverable once missiles and fuel has been expended to help it out in gunfights.) the inboard Elevon4 airbrake tends to break off from aero failure sometimes, it's not a required part so, that's fine. Assuming it's flying like that, feel free to set it up against whatever you're done tuning the MiK4 into.

The stability curve should look a bit like this if your aerodynamics are correct: http://images.akamai.steamusercontent.com/ugc/432698496208751765/38E77359324B958F056F20FD51D0351D8FF6EB44/

Buy yeah I made some tiny adjustments (just to the AI settings and dynamic deflection settings) https://www.dropbox.com/s/wb2f5mibw9e4411/F-3C%20Pidgey.craft?dl=0

Edited by Phearlock
Link to comment
Share on other sites
Phearlock

Phearlock

Posted
Posted

If you want some tips on how to improve it. I'd suggest looking at the dynamic deflection settings of the F3C's pitch authority. Usually, the aircraft would have maximum maneuverability available at low speed, and then get steadily more stable up toward the mach .8 to 1.0 region as the control surfaces become less effective. Which means you usually can't turn as hard, this is further enforced by the dynamic deflection settings that drastically reduce control authority around that region due to dynamic pressure. That's part of the main reason most fights end up with just big turning radius at high speed.

Now, notice how the F3C gets around this: it's actually only allowed 50ish% max deflection when at 0Q atmospheric pressure, which is takeoff and low speed. While once it enters a bit of atmospheric pressure, typical of low altitude flight at around .7-1.2 mach the dynamic deflection is set to provide 100% control authority to all the surfaces relating to pitch control. This compensates for the reduced control surface authority usually experienced at these speeds, and lets the aircraft pull close to it's maximum AoA around the transsonic region, which quickly converts a bunch of speed into drag and lift, letting it turn very well while also dropping airspeed to prevent overshooting (hence why setting 500 max speed for the F3C isn't as big a deal as with the MiKs, as the F3C will naturally bleed its excess speed once it starts turning, while an aircraft with standard dynamic deflection settings will be unable to pull enough AoA to start bleeding speed).

Now of course, letting the AI pull upwards of 15-17G's maxiumum at low altitude does mean I had to ramp up the wing strength of the main wings all the way to 2.0 to prevent it from tearing the aircraft apart. I daresay the F3C could indeed pull harder the settings currently allow, however, the AI doesn't know when to stop pulling and let the speed build up again. So to minimize the risk of the aircraft being stuck at 150 m/s where it simply doesn't have enough rate to dogfight effectively, the maximum deflection is tuned down just a little bit (though this is still an issue at times, and can be exploited if you wanted to design something specifically to defeat the F3C).

Of course, all this is moot if you don't have an aircraft that can pull 20-30 AoA comfortably without any risk of stalling out. I saw a lot of the MiK versions had issues with pulling beyond 15 AoA without starting to flop around pretty badly.

Link to comment
Share on other sites
Hodo

Hodo

Posted
Posted
Phearlock, how did you set the ctrl surfaces like that! I wanna have a go!;.;

The trick is simple. You have to place the control surfaces then set their settings before you move them into place.

It is tedious but very effective.

Link to comment
Share on other sites
Phearlock

Phearlock

Posted
Posted
Phearlock, how did you set the ctrl surfaces like that! I wanna have a go!;.;

It's all part of the dynamic deflection mod coupled with editing the FAR control deflection settings. I posted a link to the F3C a bit further up. Grab it and take a look at it yourself. The customizations should all be visible when you mouseover a control surface and hit K assuming you have dynamic deflection installed. (also, just right clicking and showing the FAR control surface settings to see how pitch/roll it set up.)

Link to comment
Share on other sites
Wanderfound

Wanderfound

Posted
  • Author
Posted
Assuming I can do this correctly, I might have a viable fighter. That is to say, I would be ASSUMING. Assuming makes a (donkey) out of you and me

Well, post the .craft file and we'll see. :D

--

I had a version of the MiK-4 that was matching the F-3C for turning performance, but it had some hard-to-fix surging problems that were leading to fuselage failures during dive pull-outs. Still working on it...

Link to comment
Share on other sites
Phearlock

Phearlock

Posted
Posted
MiK-6 in development...

Nice, I can see where you took the wing design from. I usually don't prefer the Mk2 fuselages for fighters as I find it a bit annoying to area rule. Though that isn't much of an issue for this challenge, seeing as we're not really going to be going very fast.

Link to comment
Share on other sites
Phearlock

Phearlock

Posted
Posted

This isn't a challenge where you have to finalize a design and stick with it until the challenge is "done" SpaceplaneAddict. Feel free to just post new iterations if you feel you need to improve your craft (which I assume you do, judging by your post).

Link to comment
Share on other sites
Halo_003

Halo_003

Posted
Posted (edited)

Is there a max G limit when turning? I'm currently pulling upwards of 15G at speed.

Also, can you post the reference craft file? I want to test mine against it before releasing. Thanks!

Edited by Halo_003
Link to comment
Share on other sites
m4ti140

m4ti140

Posted
Posted

1. Are you still taking submissions? Will you be doing another iteration of this?

2. Can you include community bugfixes for features related to this challenge?

2. Can you please, PLEEEEEEEEAASE add adjustable landing gear. It's really a pain in the ass to design fighters without it, it would be fine for everything else but fighters, especially single-engine, are a nightmare to setup undercarriage for, unless you design whole airframe specifically around incorporating stock landing gear - which may impair performance.

Link to comment
Share on other sites
Wanderfound

Wanderfound

Posted
  • Author
Posted (edited)
hi, i would like to participate, but i have a question: can i use like three airbrakes? i'll post a screen of the plane soon. and the craft of course

(btw, sorry for my bad english, i'm not english-native)

Use as many airbrakes as you like.

The restrictions are:

* Stock + BD Armory parts only.

* No torque wheels (including disabling cockpit torque).

* Jet propulsion only; no rockets, no RCS, no Vernors.

* Horizontal takeoff and landing capable; no VTOLs etc.

* Maximum of six missiles.

* Maximum of one each of chaff/flare dispensers.

* Fixed-aim guns only (no unlocked turrets).

* No excessive performance-enhancing clipping. Reasonable part clipping for aerodynamics or aesthetics is fine (e.g. burying ammo boxes inside an empty fuel tank), stacking five engines on a single node is not.

On the autopilot, set max speed to 399m/s, minimum altitude to 1,410m and default altitude to 2,000m/s. Everything else is up to you. The mods in play can be seen at http://forum.kerbalspaceprogram.com/threads/132274-FAR-Fighter-Challenge-BD-Armory-AI?p=2174355&viewfull=1#post2174355

- - - Updated - - -

Any protips for low speed stability? Above 200m/s I'm having no issues, smooth as can be, but when below that I get large scale stalling during turns and such.

There are a few tricks:

* Shift your wings backwards a bit to increase overall stability.

* Use the AoA settings on pitch control surfaces to reduce the tendency to excess AoA. Use positive AoA on control surfaces behind CoM, negative AoA on control surfaces in front of CoM.

* Use the Dynamic Deflection settings to reduce elevator input at low pressure. Hover the mouse over your elevators and hit K to bring up the menu. It'll need fine-tuning, but for starters try setting the DD authority to 50% at zero pressure and raising it by 10% for every 10kPa of pressure until it hits 100% at 50kPa.

* Use leading-edge slats set for negative AoA to substantially increase possible pre-stall AoA.

- - - Updated - - -

Is there a max G limit when turning? I'm currently pulling upwards of 15G at speed.

Pull as hard as you can; it's primarily a contest of turning ability. Just make sure you test it first to see if those 15Gs will tear your wings off when pulling out of a dive at low altitude and 400m/s.

- - - Updated - - -

1. Are you still taking submissions? Will you be doing another iteration of this?

2. Can you include community bugfixes for features related to this challenge?

2. Can you please, PLEEEEEEEEAASE add adjustable landing gear. It's really a pain in the ass to design fighters without it, it would be fine for everything else but fighters, especially single-engine, are a nightmare to setup undercarriage for, unless you design whole airframe specifically around incorporating stock landing gear - which may impair performance.

1: The contest is ongoing; the current defending champ is Phearlock's F-3C Pidgey.

2: Not sure what you're asking. If you're referring to the Stock Bug Fix mod, that was removed from play as a possible culprit in the inconsistent aerodynamics issue that caused trouble earlier.

2b: 'Fraid not. See above; we're keeping the mods as minimal as possible to avoid CPU-strain and inconsistent aero problems. If you have a look at the posted videos, you'll see an assortment of landing gear setups that work. Keep in mind that it's a turning rather than top speed contest, so a bit of extra drag from tall landing gear doesn't really hurt your chances.

Edited by Wanderfound
Link to comment
Share on other sites
Wanderfound

Wanderfound

Posted
  • Author
Posted (edited)
Well, crap. The K-16 will be anihilated, clearly. Mean people!

Time to update to the K-18, then. :)

Also: an entry doesn't have to win to be worthwhile. One-sided fights don't take long to record, and the more diversity of planes in the air the more fun it is for everyone.

If people only submit what they think are guaranteed winners the thread will stall as soon as someone posts a really good design. Instead, it would be better if everyone would put up their current best design (whether they think it will win or not), and then update it whenever they significantly improve it.

Keep the challengers coming, even if they're likely to end up as shrapnel. I can fly run-offs between the losers to create a ranking table.

Edited by Wanderfound
Link to comment
Share on other sites
Guest
This topic is now closed to further replies.
 Share
Go to topic listing
×
×
  • Create New...