How to properly setup AI approach & landing?

[Awger]

I'm working on some F-86 missions, trying to understand the "motivations" of the AI, and I'm having a problem setting up a reasonable and consistent landing scheme. In order to isolate the problem (?) I've setup a client that's just sitting on the ramp so I can watch the AI do its thing.

I have two groups of two F-86s takeoff / form up / ingress / attack / egress all working reasonably well (although I don't understand why the AI insists on doing 4G turns in the pattern, overshooting the heading to the next waypoint by 20deg, and then snaking to correct), but I'm stumped on how to setup a "proper" approach and sequencing for landing.

I've gotten things to work once or twice, but it's very inconsistent.

What I'm looking for is some general guidance on how to setup the waypoints so that two groups of two units will properly sequence for an approach and landing. Info like:

• minimum distance (time?) between groups
• allowable formations (echelon? trail?)
• target altitude / airspeed for waypoint prior to landing waypoint
• best position for waypoint prior to landing waypoint
• differences in AI behavior at different skill levels

I've noticed that at Veteran/Ace a 2-ship will do a formation landing, but at Trained they just refuse -- forcing #2 to go-around and wait for #1 to touch down before starting his approach.

Only once have I gotten both groups to do a formation landing with about 1nm separation ... re-ran the mission a second time and AI did something different.

Any suggestions on what to tweak are appreciated.

 

BombingTest3.miz

[Grimes]

Making them do a straight in approach is the most efficient way possible. The problem is if anything in the sequence interrupts the AI then they will enter the pattern behavior and will land seemingly disorganized. I think they need around 45 seconds of separation to land without entering the pattern. Placement of the last waypoint matters to the extent that the AI need to be able to make their approach safely/efficiently from that position. It is kind of like with the bombing tasks where there is an invisible line somewhere that the AI will first decide to extend or do whatever rather than directly moving toward their goal. Best to align the last waypoint with their approach and make it reasonably low altitude for the AI to directly enter their approach. 

[Awger]

A modicum of success.. I can get two pair on final with .5-.7 mile (10-15sec) separation and landed safely... any closer than that and there will be a collision somewhere on the runway (usually when one of the lead element stops short... yo dude, there's another 1,000 feet of runway out there, no need to torch the brakes).

Straight-in approach from 10 miles that puts them 180kn at 2,000ft 3 miles from threshold.  But... even setting them at the correct speed and alt they do this odd clearing turn at about 5 miles.. to dump speed and alt? even if they don't need to.

I've also been able to get two pair to land on top of each other. Not sure how I managed that, but somehow, if you get two pair close enough, they stop caring about each other and go straight for the runway.

If two pair are on approach farther apart than some distance approaching .75 mile the second element will stay on approach until its time to drop gear and then power up and go around. Minimum distamce outside of .5 mile is around 4 miles, which gives the first pair time enough to touch down before the second pair hits the gear drop distance.

Seems like the only way to get a consistent pattern is to send one pair at a time with 4 mile separation.