correct as is SEAD AI going rogue

[DaveSD]

I can't figure out why my AI Hornets flying SEAD missions either attack the closest targets or don't attack targets at all (yes, they're armed with HARMs)...

• If I set the mission as SEAD and specify targets by choosing specific radar units the aircraft go after other radars that are closer but out of the SAM's range.
• If I set the mission as pinpoint strike and specify targets by choosing specific radar units to be "bombed," the aircraft fly to the assigned waypoints and then return home without doing anything.

I've tried setting the ROE as "only designated for the target release waypoints, and I've even tried that ROE for all waypoints, but nothing seems to work. There's got to be a simple way to do this, right? It's getting to be less and less enjoyable as I spend so much time trying to fool DCS into doing what it should be doing easily. I can think of a whole bunch of things I'd like to see changed or fixed, but those will come another time.

Dave

EDIT:

I set up a test mission with one Russian SAM radar site and successfully flew the SEAD mission to make sure the radar could be targeted. Then I created a spreadsheet of literally every combination of:

• Mission type
• Task type, including two appropriate tasks when available and three when available (when three could be stacked, the same mission was flown with one, two and three tasks). In the case of multiple tasks, all possible ordering of tasks was done.
• ROE, including no ROE set at all. When ROE was specified I tried it before the tasks and after
• Weapon selected (ARM when available, guided missile when available even when ARM was available, or unspecified; for missions with ARM specified I also flew them with generic guided missiles and no missiles specified).

So again, after more than an hour of testing every combination possible was flown by the AI Hornet and none of them resulted in a missile being fired.

Any ideas what might be wrong?

 

Edited May 23, 2022 by DaveSD

[Exorcet]

My steps for setting up SEAD:

Delete the default SEAD task

Set Reaction to Threat to "Evade Fire", without this they may be overly cautious and not attack because they are too busy trying to avoid enemy radar

Near the target area set up a Race Track Orbit

Use Search Then Engage in Zone to get the AI to attack, or select specific units like you do currently. Make sure this is placed above the Orbit task so it takes priority, or else the AI will orbit and not attack.

Generally I never have to touch RoE.

Pinpoint Strike is generally for buildings and not vehicles. CAS is probably the next best alternative to SEAD, although SEAD should work fine.

[DaveSD]

I'll give that a shot, Exorcet. You're a saint!

I wish I could figure out why ED has to make things so difficult. For example, let's say I have a bunch of similar units - tanks, let's say - and I want to change all of their skill levels from excellent to random. As it is now I have to change each one individually, but it would be nice to be able to select them all using check boxes so I can make the change one time. That's just an example of something that could speed up the mission building process. I'm compiling a list of things I'd like to see incorporated. And ED, please make the Carl Vinson fully-functional. That was my ship when I was in the Navy and I'd like to fly from it. : )

Dave

[Ignition]

You need a lot of patience to do a mission. You need to learn lua also to make decent missions.

To test the mission you need to run it a 100 times. It's a pain but a well made mission can be played a few times.

AI always gives me troubles, they do wathever they want most of the times.

[DaveSD]

Hi Ignition,

I'm not the expert in terms of customer satisfaction that ED is, they get the vibe of thousands of users, but it seems to me as one guy that if realistic, complex missions are touted as a major selling point and if AI is part of that, then AI should be easier to use and work as intended. Workarounds shouldn't be needed and it shouldn't take 20 or more times as long to plan a mission (because workarounds have to be developed) as it does to fly one. I want to plan missions and fly them, and I don't want to become an armchair artificial intelligence scientist.

Think of it this way - if someone bought a new car that came with a radio as standard equipment, I don't think they'd be all that happy if it only received one station part of the time and only after spending hours trying to tune it, which only succeeded because you duct taped a weird mass of aluminum foil to the antenna. That's our AI in many cases.

The "not ready for prime time" AI can't really be taken away now, but I know that if I ran the company I wouldn't have made it available until it worked much, much better than it does now. For me, DCS is a one-against-many (me against enemy AI, which seems to actually work) sim and really not worth the money I dumped into it based on what was promised. For others, if you're happy with DCS then I'm glad that's the case.

Edited May 23, 2022 by DaveSD