Autonomous CAP and GCI AI fighter script

[Pikey]

it can break as quick as 1hr 45 mins on that version, the issue persists, if half the flight is shot down, it stalls.

http://www.twitch.tv/thebgpikester/v/52765116

just a x15 demo you can watch it on and the mission.

gcicap-testing.miz

[lukrop]

Any chance you could provide the mission in which you are experiencing the GCI spamming and CAP not respawning, Quax?

 

edit: thanks Pikey. Damn I was hoping this fixed the issue. I'm sure that I saw "half of a CAP" despawn, though.

Edited March 6, 2016 by lukrop

[Shadow KT]

Hey, here is my mission I've been working on. I moved the trigger for the spawning a bit and now some how 5/5 times they have't crashed into each other yet.

 

Also for some reason the CAP flights fly very low and do stupid shit and crash, especially the ones in the mountains. (min alt is set to 5,5 km)

 

Another thing is when a GCI or CAP flight RTBs it doesn't despawn. Only when the plane has taken some damage it despawns

 

[ATTACH]136122[/ATTACH]

Edited March 6, 2016 by Shadow KT

[Pikey]

I thought I've seen different behaviours with the static replacements, I think it may depend on some aircraft. It's been inconsistent but thats DCS.

 

Any chance you could provide the mission in which you are experiencing the GCI spamming and CAP not respawning, Quax?

 

edit: thanks Pikey. Damn I was hoping this fixed the issue. I'm sure that I saw "half of a CAP" despawn, though.

[wdrago]

1.5.3 Feedback with RC3.

 

Two Run while hosting online Run 1 : 4hours, Run2 : 3h30, CAP starting from Air (so in real time).

 

- On red side I recieve Awacs (DCS speack) alerts, one per ground radar unit having contacts. wich can turn the thinks into a spamming machine. I don't have those alerts on blue side (both use EWR 1L13 as radar unit).

 

- Altitude and CAPzone are not respected by both side, also speed (but think it s dcs issue for speed).

- don't have Roller coster flight path as on DCS 2.0

 

Question : does air traffic or Sam activity can breack CAP spawn ?

 

Why, cause I notice that on airfield proctected by Sam group(Buk or Hawk), when they come active the air traffic around airfield are put on "hold", AI flights cercling around airfield until Sam return passive (may i'm wrong). So could it be possible that CAP spawn be breack by SAM getting active on an airfield ?

 

- Blue CAP sometine get Air refuel, so I can have more Blue CAP than Expected...

- Blue CAP respawn before old one land, Red CAP spawn when old one landed... ?

 

- Generally CAP and CGI are spawning correctly, I got Far less TaxiWay Stuck for CAP and CGI than in old verison.

 

But here I got an exemple of Taxiway stuck on Nalchik (3hours after mission start (when I noticed)), A Red CAP grp got stuck on taxiway whitout any reason see attachments, So I took a red SU27 to squeeze them out, only wreck of CAP 7 and CGI 3 been clean up, but not CGI 4.

Blowing them unstuck CAP and CGI spawning.

 

Can you remind me Respawn timing for CAP and Wreck clean up ? clean up process is general or specific for each units ?

 

- Sometime a CAP or CGI grp got stuck, they unstuck by themself after 10 minutes...

 

- sometime CAP spawn on same parking spot (blowing each other).

 

- After RTB some CAP despawn some other not, all my CGI RTB have un/despawn.

 

Side Topic : I Feel that Speeding time for testing, mess AI behaving ? what you think of that ?

Edited March 7, 2016 by wdrago

[Stonehouse]

Think the roller coaster flight path is either using radar alt on the waypoint or when the script builds a waypoint it tries to figure out a height which will clear terrain at the point or perhaps a combo of the two.

So the plane will climb and dive according to the resultant waypoint heights along the path. Perhaps try raising the min alt value? The old script used a different method to build the waypoints, less efficient code wise but less prone to big alt changes. You could perhaps have less waypoints which would tend to smooth out things but also increase the risk of terrain collisions between any two points. Almost needs to work out the highest alt required along the path and climb to that soon after take off and stay there.

[Shadow KT]

I have mine at min alt of 5,5 km which is around 18K feet, and even so they are flying low to the ground around 1000 to 2000 feet and sometimes even lower and they crash.

[lukrop]

Yep saw that too. Waypoints are created using BARO alt. The altitude is randomized between min and max.

 

I noticed this behavior with some AI flights while controlling them with CA in missions not even using GCICAP.

 

To be honest I have no idea how to go about that issue. Function used for waypoint creation is here.

 

Edit: just pushed a change which should at least prevent the roller coasting of CAP flights.

 

Edit2: Oh, and I think I have a idea why the CAP flights, which lost a ship, won't despawn. For real now. :D Going to test it.

Edited March 7, 2016 by lukrop

[Pikey]

No way... :)

I took up LUA and started building an event handler to blow them up with nukes...

Yep saw that too. Waypoints are created using BARO alt. The altitude is randomized between min and max.

 

I noticed this behavior with some AI flights while controlling them with CA in missions not even using GCICAP.

 

To be honest I have no idea how to go about that issue. Function used for waypoint creation is here.

 

Edit: just pushed a change which should at least prevent the roller coasting of CAP flights.

 

Edit2: Oh, and I think I have a idea why the CAP flights, which lost a ship, won't despawn. For real now. :D Going to test it.

We tested unit ID and it never changed from landing to parking. I'm officially out of ideas as to why.

Edited March 8, 2016 by Pikey

[lukrop]

Just released RC4. Feedback very much appreciated. ;) This should at least fix the CAP despawn issue. Regarding the GCI spamming I still need to reproduce it.

[lukrop]

As far as I can tell still no luck. I'm going to add some code to clean up any collided or stuck AI. And then I'll look at this CAP despawn issue again. They probably never get their RTB order and return to their airbase by themselves. This messes up the current cleanup logic.

 

Question : does air traffic or Sam activity can breack CAP spawn ?

Could be possible but I can't really tell but after all it's still DCS. Well not the spawn but maybe the AI behavior.

 

Can you remind me Respawn timing for CAP and Wreck clean up ? clean up process is general or specific for each units ?

 

Currently there is none but it's the next thing I'll be working on.

 

Side Topic : I Feel that Speeding time for testing, mess AI behaving ? what you think of that ?

 

Yes, I thought about that too already. Might cause problems. Thinking of event handling there.

[wdrago]

Could be possible but I can't really tell but after all it's still DCS. Well not the spawn but maybe the AI behavior.

 

I made some more run of RC3. and now i'm definitely sure that A.T.C. is one of the reason (the main reason) why we get CAP and CGI stuck.

Cause I notice that stuck issue offen happen when CAP and CGI are on runway "entrance", I supposed they waiting for an ATC clearance wich never come...

 

On my map the scenario at 1 hour 30 mission run, trigger the battle of kutaisi, wich make kataisi airfield quite busy for 15min with 4 to 6 air groups taking off and landing during that time. and very offen one of my AI group "banzai" get stuck in "inboud state" waitting desperately a clerance to land, wich will never come causing lost of my brave pilots :O

As Player I also have those ATC issue when plane take off or land just before my request I have to spam clearance for take off or landing before recieving one. also read over here (104th admin of remember well) that ATC not working properly.

 

So I'm sure now, that we have an ATC issue, wich have affect on your script.

 

I gonna make a request in that way.

 

RC4 !!! Whouhou gonna tried it right now !!! ;)

Edited March 9, 2016 by wdrago

[lukrop]

RC4 !!! Whouhou gonna tried it right now !!! ;)

 

You'll have to hold back your horses for RC5 :D RC4 isn't fixing it. I'm approaching the issue now entirely different in RC5. I'll release ASAP.

[wdrago]

You'll have to hold back your horses for RC5 :D RC4 isn't fixing it. I'm approaching the issue now entirely different in RC5. I'll release ASAP.

 

you make me Hungry now :chef: !! ....

 

I reproduced the stuck issue without any script here

Edited March 9, 2016 by wdrago

[Pikey]

We made an event manager that can clean up the rogue parked aircraft that sit there for ever. However the CAP distpatcher doesn't try to spawn anything so its not waiting for them at all.

 

Might help you know where it is in code.

[Quax456]

MIST_development ERROR

 

Dear Lukrop,

 

I have a big problem with the development Version of MIST.

It crashes the Tables Mist is providing....

 

No, it is no crash, but missing or false entrys. Yea thats it :helpsmilie:

 

like mist.DBs.deadObjects

 

It causes to miss many entrys and if there are entrys they are all BUILDINGS!

 

I have testet it

[Pikey]

Mm, the mist fork is s touchy subject, i'm not keen on it as the workload to support this and a mist fork may be too much work....If I use this fork then three other scripts in the same mission then have a loss of 'support' if something goes wrong.

 

Is there anyway you can just contain and rename the functions you need and keep them in your script, rather than insisit on your version? I know we have to use yours, as normal MIST just errors this script right away.

[lukrop]

Dear Lukrop,

 

I have a big problem with the development Version of MIST.

It crashes the Tables Mist is providing....

 

No, it is no crash, but missing or false entrys. Yea thats it :helpsmilie:

 

like mist.DBs.deadObjects

 

It causes to miss many entrys and if there are entrys they are all BUILDINGS!

 

I have testet it

 

I investigated and found out why mist.DBs.aliveUnits and in succession mist.DBs.deadObjects contained invalid/wrong data and just sent a pull-request to upstream.

 

Thanks for spotting that nasty bug.

 

Mm, the mist fork is s touchy subject, i'm not keen on it as the workload to support this and a mist fork may be too much work....If I use this fork then three other scripts in the same mission then have a loss of 'support' if something goes wrong.

 

Is there anyway you can just contain and rename the functions you need and keep them in your script, rather than insisit on your version? I know we have to use yours, as normal MIST just errors this script right away.

 

I'm using/requiring the official development version of MIST from Grimes' github repository.

 

Since it is MIST's development version I guess sooner or later Grimes will merge the changes back into master and release a new MIST version.

 

Unless Grimes makes a 180° turn and drops all the changes (already committed to the development branch of his repository) I don't think you have to fear anything regarding support.

 

Working on GCICAP I found some bugs, mostly in code contributed by me to MIST, which I fixed and sent pull-requests. Therefor I'll keep using the MIST development version, including bug-fixes I made, until Grimes merges them into his development branch.

 

If you are interested you can always have a look at the currently pending pull-requests of MIST.

 

Regarding GCICAP progress: I wrote a garbage collector which is working pretty well but there still are cases where GCICAP looses track of groups and is not able to remove them correctly, leaving zones "guarded by ghosts" and therefor not spawning any new CAPs. Still working on it, but I think I'll need a break for some days because I can't wrap my head around it anymore. :joystick: :music_whistling: But I'm staying positive, we'll fix it. ;)

[wdrago]

Regarding GCICAP progress: I wrote a garbage collector which is working pretty well but there still are cases where GCICAP looses track of groups and is not able to remove them correctly, leaving zones "guarded by ghosts" and therefor not spawning any new CAPs. Still working on it, but I think I'll need a break for some days because I can't wrap my head around it anymore. :joystick: :music_whistling: But I'm staying positive, we'll fix it. ;)

 

 

Take an Ice Cream !! you deserve it !!!

 

[Stonehouse]

Mm, the mist fork is s touchy subject, i'm not keen on it as the workload to support this and a mist fork may be too much work....If I use this fork then three other scripts in the same mission then have a loss of 'support' if something goes wrong.

 

Is there anyway you can just contain and rename the functions you need and keep them in your script, rather than insisit on your version? I know we have to use yours, as normal MIST just errors this script right away.

 

Better yet - report the issue to Grimes and how you resolved it and more than likely he'll incorporate it into the base Mist build. One thing we really don't want in my honest opinion is 50 different flavours of the community base scripts - think CBA in Arma3 it's big strength is there is only one CBA version current not 10.

[Grimes]

Not sure whether I'm more annoyed that I missed the error when I checked the code or the fact that it was caused by renaming local values when it wasn't really needed to. Anyways the pull requests were merged.

[Stonehouse]

1.5.3.1 Changelog

Crash when attempting to call Object.destroy() in landing event handler is fixed.

 

So you no longer need to schedule the despawn - wonder if that may solve some of your issues?

Edited March 11, 2016 by Stonehouse

[lukrop]

I finally, actually fixed the despawn/respawn issue. Logic fail on my side. After the first removal of some flights the table, containing all GCICAP flights gets "reorderd" by LUA (using table.remove()). I still used the old index to remove flights and that's where bad things started to happen. Removing flights which where actually still in the air.

 

I'm testing a bit more and I'll push a hotfix. I'm working on making the CAP orbit using a race-track pattern and have a customizable vul time. I tripple checked the waypoint altitudes. For some odd reason they start randomly diving on some waypoints. Using race-track pattern orbit is way more realistic and will hopefully fix this roller-coasting for good.

 

Stay tuned! ;)

[Pikey]

Can't wait and I'm really pleased there's some good collaboration guys :) Waiting with baited breath that this is the one. The racetrack is a beautiful bonus if you can get that working. Thanks guys!

Lukrop can you link to GCICAP github yet in your signature, I constantly hgave to look through places? :)

[Quax456]

@Lukrop

Sounds good my friend....