Consistent AI SEAD Behaviour - ACHIEVED!!

[Fubarbrickdust]

So basically I've been trying to make a multi package strike mission consisting of SEAD, STRIKE & CAP elements...cool, good for me.

 

However, I was stopped dead in my tracks by the suicidal/dumb/frustrating AI behaviour that would either do nothing, not attack the specified target types, drop their junk or just plain wade in and get themselves killed on a suicide mission.

And also NEVER using their Anti-Radiation Missiles (AGM-88C) at anywhere near RMAX, and never at the right targets!

 

I had a search on the issue and it appears I'm not alone...

 

 

 

 

But I seem to have stumbled upon task & trigger logic that's getting good results! It's a bit of a cheat - but nowhere near as much as just triggering the units to explode on a given cue.

 

All testing so far done with the SA-10 SAM Site template, which as far as I can tell has the longest acquisition reach (it's got the biggest circles around it, is what I gather)

 

There's a few requirements, but they're basic.

 

1 Trigger Zone

• Setup to just encompass the SAM Site engagement radius.
  

 

2 Correctly placed waypoints

• 1st WP just in the border of the engagement range/trigger zone.
  
• 2nd WP at a safe distance outside.
  

 

2 Triggered Actions (AI PUSH TASK)

• 1st action setup to 'SEARCH THEN ENGAGE IN ZONE' - with the radius centred on and small enough to cover your enemy SR & TR units, nothing else (The ME ignores any attempt to specify SR/MR etc so I wouldn't bother).
  
• 2nd Action setup to trigger a 'SWITCH WAYPOINT' command to a waypoint outside of the SAM engagement zone.
  

 

2 Triggers (to push the actions)

• One to trigger the 'SEARCH THEN ENGAGE IN ZONE' on a condition of your choosing.
  
• Another to trigger the 'SWITCH WAYPOINT' action when your SEAD flight enters the Trigger Zone (and by proxy, the SAM engagement radius).
  

 

NOTE: This whole thing appears to work without having to force a load of ADVANCED(WAYPOINT ACTIONS) to get the SEAD flight to chill.

 

An appropriately set up SEAD package - VERY IMPORTANT

This made all the difference for getting a reasonable RMAXish weapon release.

 

• A flight of F-18C's (just the standard ones, not the lot20's) loaded with 4 AGM-84E & Lightning Pod each. (forget the AGM-88's, no AI will let them go until they're as good as dead!).
  

 

VARIATION & TWEAKING

Altitude & speed play a large factor. I set mine to 25000ft@400 and got consistent successful hits every time. Lower & slower mixed it up a bit.

 

Placement of the 1st waypoint (the one just inside of the trigger zone).

This is the cool part - If you're just on the cusp or very shallow in, they'll turn out with plenty of space. The deeper into the zone you put it, the more it feels like you've got cocky jockeys! I know the're being overridden by the AI PUSH TASK to SWITCH WAYPOINT but they go just that bit deeper into the zone before turning out, and give the SAM a chance to fire off a few. Take the WP all the way to the site and it starts to push the luck!

 

So far, I've only tested with the SA-10 site & the AI F/A-18C, with the most effective load-out for stand off engagements being the one mentioned...but the results are consistent, and most importantly effective at creating a successful SEAD mission.

 

See screenshots and attached .miz file.

 

I've let the mission file play out dozens of times with different little tweaks and seem to consistent/predictable results. Tested on PG also.

AI SEAD TEST.miz

Edited March 10, 2019 by Fubarbrickdust

[Habu_69]

Good stuff. I have had reasonable success with using F-16 Bl 50 SEAD on the shorter range SAMs (SA-15 for example), but not so much on SA10, SA-11.

[enigma6584]

Interesting indeed! I will have to experiment with this myself. Thanks for sharing.

[Ghostrider142]

I haven't touched DCS since spring of last year, I was hoping to come back (Tomcat release) and find ME having been given some TLC. Such massive potential in this editor just by simply fixing the erratic AI!

Good job and good luck getting it to work every time

[Fubarbrickdust]

snip...

Good job and good luck getting it to work every time

 

Obviously I can’t and won’t speak for the manipulation of the AI in general, but for this particular use case and purpose (less suicidal and more realistic standoff engagement range for SEAD/DEAD) the technique used is consistent and very repeatable. I spent about 4hrs testing it with different variables and locations.

 

It’s irksome that the AI can be as troublesome as it is, but I can imagine coding smart AI that works predictably in every scenario is no small feat. And I’m sure it’s on the endless todo list that ED has.

 

It shouldn’t be this way, but sometimes the gratification I get from DCS is finding a way around or alternate solution for the engine limitations or quirks (read: bugs). Some of which are pretty frustrating, plain and simple mind you. I’ll totally concur!