TimC340 wrote: ↑27 Jan 2021, 12:23
I remember him talking about Innsbruck when explaing how to make approaches to avoid terrain - fortunately, that expertise is still all available in the archives at FDev.
For AI (at least in FS9) to make an angled approach it boils down to creating just one approach leg, by setting an Initial Fix where you need it to be, and picking a point along the runway's extended centerline where you want the AI to intercept the final approach path. AI on approach is always "sniffing" for that extended centerline and, when closing in on it will execute a turn onto final that doesn't need to be programmed since it's just what AI does, a bit like a "homing instinct". For my Battersea AFD I have the AI helos descending along the Thames from either direction and all that's needed are tracks from an Initial Fix to cross the extended centerline, where I placed a FF (Final Fix) to make reading off headings simpler (just move the fix and read a revised heading, and enter that into the leg's heading field). For slow aircraft such as helicopters this intercept point can be very close to the threshold, while for airliners it's a whole other problem with the higher speeds, especially if the closing angle is quite sharp, at it is for the offshore arrivals into Nice/LFMN - or the right-angle turns to final at Abu Dhabi's helicopter runway. ADE's graphic interface makes this phenomenally simple, a world away from having to write the XML and compile it separately.
And obviously this fake ILS is only used if the AI is flying IFR in the first place.
But an angled leg from an IF waypoint that crosses the extended centerline is all you really need for AI purposes if they're avoiding terrain or even on noise abatement procedures etc.
TimC340 wrote: ↑27 Jan 2021, 12:23
When I get a chance, I'll go back through some of the overlays I've done and see if this radio effect is at play, and how I've managed to inadvertently avoid it causing problems. My efforts have always focussed on avoiding AI jumping networks. I'd never really worked in FS9, having made the move to FSX on Day 1 and subsequently to P3D, until very recently when helping (or trying, at least) JY out with some of his stuff. There are some small differences, I believe, in how AI is handled in FS9 from the later sims, but I wonder if this is one of them?
If an AI arrival gets its parking allocation at a hold-short node after arrival my guess is that it should work its way through the network without any problem, at least, so far as I can tell. The problem at Aberdeen was that most arrivals vacated along a helicopter runway (in either landing direction, a different helo runway for each direction), so my trying that particular layout was asking for trouble, since it could catch an arrival on the other AFD quite easily. Especially since it was my first go at an overlay even though I've been working with AFCADs for a long time. You think you've got all the rules down, and then along comes an overlay to change all the rules.
They're basically just equal priority scenery areas that interfere with each other invisibly, and you end up reduced to reading signs (like the ATC window) to attempt to figure out what the AI is doing, or why it isn't!
None of this is documented in SDKs etc, of course.
Differences in AI between sims is unknown to me. I have FSX but never used it. It's such a rudimentary implementation that I'd be surprised at any significant differences. The most seismic one in the range of FS environments for me (FS5.1 thru FS9, so several pre-AI versions as well) was the loss of 1-way runways between FS2002 and FS2004/9, so for me it's not always been a path of improvements! AI has always been the poor relation in terms of implementation and support, which makes it all the more amazing that so many users manage to wring what they do from it.
Back on topic after that epic attempted thread hijack - I think from what I've seen so far that it is proximity to a runway in an overlay that causes the radio frequency masking, and that it doesn't seem to matter whether the overlay shares r/t frequencies with the co-located airport, or where the reference point is (I moved that around and it made no difference). Whether it's the actual defined runway, the black runway links, hold-short nodes, or even the start points is an open question. I personally think it's the defined runway(s) in the different AFDs, but I could easily be mistaken, not having any other overlays to my name as yet.