[ASC-III] Air Superiority Challenge - King of the Hill (BDArmory 4v4 AI Duels: WW1 Theme) - Now Concluded!

[drtricky]

@JollyGreenGI Creating a fifth aircraft with no weapon manager or afterburner and having it hover seems to make the results more random, although I don't like the view of the battle it offers. I also don't like how it makes laser battles essentially a mere game of chance without armor.

If you ask me though, it's just too much trouble making laser-laser battles fair

Edited April 24, 2016 by drtricky

[JollyGreenGI]

@drtricky You would have the fifth craft follow one of the flights each round, maybe with the AI pilot. And yes, armour is something that will have to be developed against lasers.

[SuicidalInsanity]

1 hour ago, drtricky said:

 I'm noticing something very troubling about laser-laser battles in 1.1. I've noticed that if both sides have ships with potent lasers that engage each other, the side that wins is the side who has the camera focused on them. I fought one of my laser ships against solarius III, and I noticed that in the laser on laser engagements, the side that won was always the side that had the camera focused on them. The side that had the camera focused on them was always the first to fire, and thus the first to kill.

Could you or anyone reading this possibly confirm this for me?

Fairly certain this isn't the case
I was curious, so I decided to do Science! at the problem, testing methodology below. 40 test battles later, and it seems that camera focus is irrelevant in the outcome of laser v laser battles. The results of Tests 6 and 8 suggest craft position as a result of pre-competition maneuvering into position has an effect on the outcome of laser duels. I suspect craft alignment and orientation when craft are spawned in on the ground may also be a factor  One thing I noticed was each fight was preceded by an identical takeoff sequence - every time the craft seemed to position themselves in the same point and orientation in the sky at the start of the competition, which would explain near identical results and and battle damage each match. Curiously, however, even when randomizing craft position at competition start in Test 7 and 8, kills continued to be streaks rather than something closer to random, which suggests either Test 7 and 8 attempts to randomize craft position in the air were insufficient, or there is some variable that's been overlooked.

Testing methodology:

Spoiler

In an attempt to remove extraneous variables, all test duels were 1v1 matches between two unmodified Solarius III's using the BD competition mode at a default 8000m

Hypothesis 1: camera-focused vessels win laser duels.

Spoiler

Test took place at the KSC. First test A team Sol (prime)spawned on the runway, B team Sol (secundus)spawned on the grass adjacent. Camera was focused on the A team Sol for the duration of the fight. A 1v1 duel was fought 5 times, all five times the camera focused A team Sol won.

Second Test: setup was identical to the first. B team Sol was camera focused for the duration of the fight. A 1v1 duel was fought 5 times, all five times the camera focused B team Sol lost

Third Test::Setup was identical to Tests 1 and 2. Camera was focused on A team Sol during takeoff, and switched to B team Sol at competition start. a 1v1 duel was fought 5 times. All five times the B team camera focused Sol lost.

Hypothesis 2: Runway height gives unfair advantage

Spoiler

Test 4: Test took place adjacent to KSC with both team Solarius III's spawning back to back. Camera was focused on A team Sol for the duration of the fight. Five 1v1 matches fought, A team Sol won 5 times

Test 5: Test setup identical to Test 4. camera was focused on B Team Sol for duration of fight, 5 1v1 matches fought, B Team Sol lost 5 times

Hypothesis 3: team alignment has a factor

Spoiler

Test 6: Setup was identical to Tests 4 and 5. Following setup, the Sol prime (A team) was assigned to the B team, and vice versa. Camera was focused on Sol prime (B team nee A team) for the duration of the fight. 5 1v1 matches fought, Sol secundus (A team nee B team)lost 4 times, Sol prime lost 1 time - this is where things get interesting. The match where Sol secundus won was glitched - the competition mode sent the planes toward each other, instead of away from each other; competition mode had to be toggled off, then back on for fight to progress, resulting in a Sol secundus win.

Hypothesis 4: craft aerial position has a factor

Spoiler

Test 7: Previous tests the maneuvers the Sols used to get into position during competition setup were identical. Each match the camera focused craft would make the exact same pre-match  maneuvers. Setup was identical to tests 4 and 5. 5 1v1 matches were fought. matches were initiated by setting both Sols to A team, and activating pilot and guard modes to get Sols to takeoff and go into a holding pattern, one Sol was transferred to B team, and the competition mode activated. Camera was focused on A team Sol for duration of fight. A team Sol won 5 times.

Test 8: setup and procedure was identical to test 7, except camera was focused on B team Sol. 5 matches were fought, A team Sol won 4 times, B team Sol won 1 time. Again, the one win was when I took slightly longer to set a B team Sol and activate competition mode

Conclusion: Solarius prime scored the vast majority of kills regardless of which craft the camera was focused on, start location, competition start location when airborne. or team affiliation. Some unidentified factor is at work here, further testing required.

The results as is are somewhat troubling. The next step would be to repeat these experiments with two different laser craft and see if similar results occur.

[Hobbes Novakoff]

13 hours ago, SuicidalInsanity said:

Fairly certain this isn't the case
I was curious, so I decided to do Science! at the problem, testing methodology below. 40 test battles later, and it seems that camera focus is irrelevant in the outcome of laser v laser battles. The results of Tests 6 and 8 suggest craft position as a result of pre-competition maneuvering into position has an effect on the outcome of laser duels. I suspect craft alignment and orientation when craft are spawned in on the ground may also be a factor  One thing I noticed was each fight was preceded by an identical takeoff sequence - every time the craft seemed to position themselves in the same point and orientation in the sky at the start of the competition, which would explain near identical results and and battle damage each match. Curiously, however, even when randomizing craft position at competition start in Test 7 and 8, kills continued to be streaks rather than something closer to random, which suggests either Test 7 and 8 attempts to randomize craft position in the air were insufficient, or there is some variable that's been overlooked.

Testing methodology:

  Hide contents

In an attempt to remove extraneous variables, all test duels were 1v1 matches between two unmodified Solarius III's using the BD competition mode at a default 8000m

Hypothesis 1: camera-focused vessels win laser duels.

  Hide contents

Test took place at the KSC. First test A team Sol (prime)spawned on the runway, B team Sol (secundus)spawned on the grass adjacent. Camera was focused on the A team Sol for the duration of the fight. A 1v1 duel was fought 5 times, all five times the camera focused A team Sol won.

Second Test: setup was identical to the first. B team Sol was camera focused for the duration of the fight. A 1v1 duel was fought 5 times, all five times the camera focused B team Sol lost

Third Test::Setup was identical to Tests 1 and 2. Camera was focused on A team Sol during takeoff, and switched to B team Sol at competition start. a 1v1 duel was fought 5 times. All five times the B team camera focused Sol lost.

Hypothesis 2: Runway height gives unfair advantage

  Hide contents

Test 4: Test took place adjacent to KSC with both team Solarius III's spawning back to back. Camera was focused on A team Sol for the duration of the fight. Five 1v1 matches fought, A team Sol won 5 times

Test 5: Test setup identical to Test 4. camera was focused on B Team Sol for duration of fight, 5 1v1 matches fought, B Team Sol lost 5 times

Hypothesis 3: team alignment has a factor

  Hide contents

Test 6: Setup was identical to Tests 4 and 5. Following setup, the Sol prime (A team) was assigned to the B team, and vice versa. Camera was focused on Sol prime (B team nee A team) for the duration of the fight. 5 1v1 matches fought, Sol secundus (A team nee B team)lost 4 times, Sol prime lost 1 time - this is where things get interesting. The match where Sol secundus won was glitched - the competition mode sent the planes toward each other, instead of away from each other; competition mode had to be toggled off, then back on for fight to progress, resulting in a Sol secundus win.

Hypothesis 4: craft aerial position has a factor

  Hide contents

Test 7: Previous tests the maneuvers the Sols used to get into position during competition setup were identical. Each match the camera focused craft would make the exact same pre-match  maneuvers. Setup was identical to tests 4 and 5. 5 1v1 matches were fought. matches were initiated by setting both Sols to A team, and activating pilot and guard modes to get Sols to takeoff and go into a holding pattern, one Sol was transferred to B team, and the competition mode activated. Camera was focused on A team Sol for duration of fight. A team Sol won 5 times.

Test 8: setup and procedure was identical to test 7, except camera was focused on B team Sol. 5 matches were fought, A team Sol won 4 times, B team Sol won 1 time. Again, the one win was when I took slightly longer to set a B team Sol and activate competition mode

Conclusion: Solarius prime scored the vast majority of kills regardless of which craft the camera was focused on, start location, competition start location when airborne. or team affiliation. Some unidentified factor is at work here, further testing required.

The results as is are somewhat troubling. The next step would be to repeat these experiments with two different laser craft and see if similar results occur.

Hmmmm... Interesting. Maybe try different competition distances? Or turn on Peace Mode, make everyone take off and enter a holding pattern, then turn peace mode off.

[inigma]

Thanks for testing various configs with the lasers. Be sure to feedback your concerns and findings to @BahamutoD.

The ASC Host team has been aggressively testing recompiled BDArmory builds, and I think after we finish the current ASC-I roster, we will be able to open up ASC-II immediately after now that BDA is updated for 1.1 officially.  Anyone who would like to join the match hosting team for ASC and its divisions, let me know!

[inigma]

 

@Monkey29399, @Kokanee @Dman979 @JollyGreenGI @GDJ

Last call to get your fighters in. We will be completing the roster for ASC-I this week.

Edited April 25, 2016 by inigma

[Dman979]

20 minutes ago, inigma said:

Thanks for testing various configs with the lasers. Be sure to feedback your concerns and findings to @BahamutoD.

The ASC Host team has been aggressively testing recompiled BDArmory builds, and I think after we finish the current ASC-I roster, we will be able to open up ASC-II immediately after now that BDA is updated for 1.1 officially.  Anyone who would like to join the match hosting team for ASC and its divisions, let me know!

First, thanks for the ping.

Second, does Adjustable landing gear work in 1.1?

[inigma]

1 minute ago, Dman979 said:

Second, does Adjustable landing gear work in 1.1?

Not yet.

[Dman979]

Just now, inigma said:

Not yet.

Hmm, well, that's a problem, because some of my fighters require it to take off.

[JollyGreenGI]

1 minute ago, inigma said:

Monkey29399, Kokanee, Dman979, JollyGreenGI, GDJ

Last call to get your fighters in. We will be completing the roster for ASC-I this week.

I posted a few days back an additional fighter to compliment Fiesta QUIET, it should be in the ASC hangar on KerbalX.

 

1 minute ago, Dman979 said:

Second, does Adjustable landing gear work in 1.1?

They can be used as landing gear, but the adjustability is non-functional.

22 minutes ago, inigma said:

 Anyone who would like to join the match hosting team for ASC and its divisions, let me know!

I think my rig is decent enough to run matches, especially with 1.1 officially out. And I just got out of exams, so I'm pretty free for now.

So, sign me up.

[boomchacle]

in 1.1 I hope my fighter isn't going to lag your computer...

[GDJ]

34 minutes ago, inigma said:

 

@Monkey29399, @Kokanee @Dman979 @JollyGreenGI @GDJ

Last call to get your fighters in. We will be completing the roster for ASC-I this week.

I'm good. Run mine as is. Whatever happens I'm good.

And I won't be coming back with KSP 1.1 yet. It's a complete fart-up with Mac OS X. Horrid SAS issues, landing gear (I think I have that fixed) but after 3 minutes of level flying, the SAS completely loses it's marbles and shoots the plane straight down and it's uncontrollable.

Edited April 25, 2016 by GDJ

[SuicidalInsanity]

Did more testing with lasers today, and I have some good news. Today's test was an extension of the previous test, this time between a Solarius III and a new laser armed interceptor I designed primarily for speed and maneuverability; despite having fewer lasers (4 compared to he Sol's 6), and regardless of camera focus, the more maneuverable plane won, which suggests that maneuverability and/or speed is still a relevant design concern for laser craft.

Comparing the fights, best guess is the AI gets wonky when two identical planes are flying against one another; each are following an identical AI moveset, so if one plane starts out with a minuscule advantage due to positioning, it will win until that advantage is removed. 

I also tested non-laser armed crafts against a Solarius to work out design strategies for reducing to outright negating the threat from laser craft. Amusingly lasers seem to be easier to defend against than AIM-120 spam, and that anti-laser measures (ablative radiator plates/ Cockpit & AI module location/ overall craft profile) can fairly easily be integrated into conventional designs without sacrificing flight characteristics.

[inigma]

Next match.

 

[JollyGreenGI]

@inigma Direct links to my current submissions:

https://kerbalx.com/JollyGreenGI/TAM-Y22D-Fiesta-QUIET
https://kerbalx.com/JollyGreenGI/TAM-Y22E-Fiesta-RIOT

[g00bd0g]

15 minutes ago, SuicidalInsanity said:

Did more testing with lasers today, and I have some good news. Today's test was an extension of the previous test, this time between a Solarius III and a new laser armed interceptor I designed primarily for speed and maneuverability; despite having fewer lasers (4 compared to he Sol's 6), and regardless of camera focus, the more maneuverable plane won, which suggests that maneuverability and/or speed is still a relevant design concern for laser craft.

Comparing the fights, best guess is the AI gets wonky when two identical planes are flying against one another; each are following an identical AI moveset, so if one plane starts out with a minuscule advantage due to positioning, it will win until that advantage is removed. 

I also tested non-laser armed crafts against a Solarius to work out design strategies for reducing to outright negating the threat from laser craft. Amusingly lasers seem to be easier to defend against than AIM-120 spam, and that anti-laser measures (ablative radiator plates/ Cockpit & AI module location/ overall craft profile) can fairly easily be integrated into conventional designs without sacrificing flight characteristics.

what are " ablative radiator plates"?

[JollyGreenGI]

Just now, g00bd0g said:

what are " ablative radiator plates"?

I would assume covering a craft with radiator plates that will be burnt off before vital parts of the aircraft are destroyed.

[SuicidalInsanity]

Pretty much, although covering the whole plane is a bit overkill. Test designs I tried used a trio of the small radiators bolted onto the nosecone, which was enough for them to get to gun range with minimal damage.

 

[Hobbes Novakoff]

Also, PSA regarding landing gear: The stock medium landing gear is now a large airliner landing gear, so you will need to tweak your wheel setup. (Also, turn off rear wheel steering as it appears to be on by default.)

Otherwise, it doesn't seem that any other changes are needed. I've been doing plenty of agressive testing of planes in 1.1 (a sample battle featuring some of @JollyGreenGI's Matadors can be found here) and performance is excellent thanks to multicore physics. We're also thinking of bumping the match size up to 3v3, but no guarantees.

[JollyGreenGI]

@Hobbes Novakoff My god, that battle was very well put together. I'm not gonna point out any particular part, because damn, you gotta see it to believe it. I'm glad someone's still getting some use out of the Matadors.

[Monkey29399]

Here's my more improved craft:

https://www.dropbox.com/s/gdnymv35ufle9no/Fighter%20Mk1-2.craft?dl=0

I would also like this craft to replace my old fighter Mk1.

 

Also, I have a question.

Since BDA doesn't activate the jammers automatically, do you activate them manually at the start of the match?

Edited April 25, 2016 by Monkey29399

[Hobbes Novakoff]

10 hours ago, JollyGreenGI said:

My god, that battle was very well put together. I'm not gonna point out any particular part, because damn, you gotta see it to believe it. I'm glad someone's still getting some use out of the Matadors.

Thank you!

[Jakerblam22]

A good match (at least for you), @SuicidalInsanity! That laser cluster (I'm assuming you clipped a bunch together b/c one laser doesn't kill the Nightmare like that) really ripped me apart! Nice job there. Thanks to @inigma for hosting the match. I've got something else in the works for ASC II, though.......hehe.

[Spearka]

okay, i feel better now. I've modified the Thassa fighter to include that Kerbal seat and also added 2 AI cores on each engine block considering each half performs too weirdly. I also added struts on the front as in 1.1 the front wiggles around a bit. Otherwise it's unchanged.

Yes I am re-entering if anyone asks

KerbalX Mirror

Edited April 25, 2016 by Spearka

[inigma]

Just now, Spearka said:

okay, i feel better now. I've modified the Thassa fighter to include that Kerbal seat and also added 2 AI cores on each engine block considering each half performs too weirdly. I also added struts on the front as in 1.1 the front wiggles around a bit. Otherwise it's unchanged

KerbalX Mirror

Do you want me to run it in ASC-I? If so, I'll need a KSP 1.0.5 saved version. If not, hang on to it for ASC-II starting sometime next week.