AI Aircraft SID and STAR Controller

[ysfsim 22]

I noticed that sometimes autogen approach are saved in the autogen cached.folder in aisidstar folder. Might be a good start fir further.customization. Is it so intended Roland? However it seems that the file gets deleted once the program exits. I couldnt see the file at all times either when running the program.

Couldbyou shed some liggt on this?

Thank you!

wow that is odd. Once the program forms an autogen approach its sup post to stay in that folder. The reason is that it takes time for an autogen approach to form. Notice that thary are formed when air are taking too long to land. In the latest version, Roland decided to have the autogen approaches stored rather than being created every session. Make sure you have the most recent version of aisidstars. Not sure why yours are disappearing but they stay on mine.

[FlyTweety 2]

Well, finally, to answer the question i decided that i will write custom approach files for each Airport of my interest (or there will be downloadable procedures one day).

Until today i´m not really satisfied with autogen approaches, they work to accidently for me.

But, i´m convinced that a lot of the speed adjustments belongs to the separation.

Because, what else can be thereason to slow down to 47 kts with an B757-300?

It is lot more than slowing down to Flap speed. And that is what is disturbing a lot.

Might be that the Traffic have their seperation of 4 nM in my case at the begin, but slowing down that hard end up in loss of separation in seconds for all the Traffic in the final.

Sure, it should be ajustable, but for some reason it don´t work for me frequently and fall below the setting of factor 0.8

 

 

Nevertheless, also the custom Approaches have still their limits.

 

E. g. LOWI (Insbruck Austria)  RW 08 is currently not doable.

I must lead the AI to an area in the west of the Airport to get the Runway assigned. Then back to NDB RTT and further the "visual" Approach.

Problem is, if i put all waypoints in the STAR, the Approach looks fine and smooth til the end of the STAR. But instead of proceeding to the Final the AI turns left into the Mountains, obviously to catch the P3d/FSX IAF.

When i put all the Waypoints in the Valley into the Final instead, i have again the behaviour that the AI turns outright like slewed.

 

So, maybe it is possible to create a third way by using a "special" Final with the algorythms like in the STAR.

Currently i´m not able to check it out because i rearrange my Computer.

[ysfsim 22]

 

 

Great, I like 10000 too (I want to keep the enroute traffic density normal). If you see a few AI at an airport that shouldn't have any AI traffic, chances are those AI are departing from or arriving from an airport on the "do not delete AI" list.

Now that I understand that, I see that it works fine now.

[bib974 0]

Indeed, I will have to check again for the stored autogen approach. I do have the latest version but not sure to have tried it yet for a long enough time.

 

FlyTweety, for LOWI you may try to position the Iaf if and faf fix for the approach somewhere more appropriate using ADEX.

 

To go back to autogen approach, i still have issues once an autogen approch is recorded and stored. Many AI planes will fly the final towards the airpot flying right and left in wide.turns, doing this again on the airport ground.

Am I the only one to have that?

 

Thank youbfor your help

Edited for typos du to smartphone typing

[FlyTweety 2]

FlyTweety, for LOWI you may try to position the Iaf if and faf fix for the approach somewhere more appropriate using ADEX.

 

 

I tried it, but without success.

Though i changed some Waypoins (one per test of course) from type "Named" to "IAF" and "FAF", P3d doesn´t care about it.

I could find the area, where i get the runway assignement for sure, but that was it.

The rest is in god´s (ah, P3d´s) hands.

[bib974 0]

I could find the area, where i get the runway assignement for sure, but that was it.

The rest is in god´s (ah, P3d´s) hands.

 

That's a good start too me! I wish I could say the same for FSX Style a mystery for me where this area is and how it is defined

[FlyTweety 2]

For LOWi basically this is easy: You need a second program like Flightsim Commander that shows you all the traffic.

 

Without AISIDSTAR running you open the ATC Window and watch/listen where the ATC assign the Runway to the AI.

Finaly you add a Waypoint (or use a existing) in this area and add it at first to the STAR´s.

 

The problem is that is looks stupid regarding traffic that is is injected in the odd corner.

The AI appears east from RTT, flys direct to the clearance area southwest of the Airport, then all the way back to RTT and starts his "visual" approach.

But i didn´t found another way to be sure that i have a valid assignement :(

[ysfsim 22]

For LOWi basically this is easy: You need a second program like Flightsim Commander that shows you all the traffic.

 

Without AISIDSTAR running you open the ATC Window and watch/listen where the ATC assign the Runway to the AI.

Finaly you add a Waypoint (or use a existing) in this area and add it at first to the STAR´s.

 

The problem is that is looks stupid regarding traffic that is is injected in the odd corner.

The AI appears east from RTT, flys direct to the clearance area southwest of the Airport, then all the way back to RTT and starts his "visual" approach.

But i didn´t found another way to be sure that i have a valid assignement :(

Why do we bother? Can't some one just make a new atc for fsx that includes all traffic (user and ai). Yes I am aware of one program that does it ( I won't name it), but it has fundamental problems that ruins the experience. It assigns sids on clearance and ai follows the stars and supposedly does holds. The real thing it does that makes it stand out from all the other atc programs out there is its ability to give taxi instructions to user and ai. If we get a program the can make the ai taxi, then this program would work without issues. It feels like we ( the flight sim community) are tip toeing around this sleeping monster called fsx. Are we afraid it will bite us? While I'm at it, I will rant some more.

Programs like aisidstars, aicull should have been around since nearly the beginning (fs9 days). It seems that a frustrated few can only be the trailblazers. This program I bet came about because some could not stand any longer watching default go arounds and straight in approaches. Your telling me these "great developers" could not think of a way to fix or work on a way to control ai? :mad:  What makes it worse is that they have the most knowledge of the fsx workings and the resources to make it happen but won't. Its not like people won't buy it, but I digress.

 

Keep up the good work Roland and keep trailblazing.

BTW anything new in the next version?

[Clipper Ocean Spray 189]

E. g. LOWI (Insbruck Austria)  RW 08 is currently not doable.

I must lead the AI to an area in the west of the Airport to get the Runway assigned. Then back to NDB RTT and further the "visual" Approach.

Problem is, if i put all waypoints in the STAR, the Approach looks fine and smooth til the end of the STAR. But instead of proceeding to the Final the AI turns left into the Mountains, obviously to catch the P3d/FSX IAF.

When i put all the Waypoints in the Valley into the Final instead, i have again the behaviour that the AI turns outright like slewed.

 

 

It is doable and I just did it.

 

http://www.mediafire.com/download/cb68q9o43d97uet/LOWI.zip

 

It is a bit odd that people seem to be gravitating toward using this program for the most extreme approaches.  Maybe I should rename the program AIExtremeApproachController?  Let's see, I've done KDCA River Visual 19, KJFK Parkway Visuals 13L/13R, Toncontin MHTG RNAV 02, and now LOWI Visual Rwy 8.  

 

The trick is to add AIImmediateForceLandAll = 1 to your .ini.  Oh right, that would be too easy.  Watching the AI taxi to the gate after landing (because AIImmediateForceLandAll deletes the AI after rollout) is just as important and enjoyable as the AI flying published procedures to the letter - right?   :rolleyes:  So, I guess we can't use AIImmediateForceLandAll after all, we'll need to engineer everything around hitting that FSX IAF.

 

So...doing it the "hard way", the first trick is to begin funneling the AI north-to-south (or south-to-north) well away from the airport so the traffic looks better over the airport (especially AI approaching from the east).  I've written the STAR to begin at the LEMSO airway waypoint, which is about 20 nm north and slightly west of the airport.  The STAR then funnels the AI to KTI at flight level (note: this isn't an official STAR entry point, but the AI actually hasn't begun the official STAR yet and moreover the AI is at flight level and thus not noticeable).  From their, the AI exits the STAR and intercept the FSX IAF for runway 8 without too much course change (be sure to add AICircleUntilAssignedRunway = 0 to your .ini for best results).

 

After intercepting the IAF, the final approach file is loaded, which is actually a combination STAR/final approach.  The AI first joins the STAR at INN, then RTT, then begins the final approach portion by implementing a procedure turn back to RTT to intercept the 210 radial, then intercept the localizer at 255, then the turn to 230, base, the 180 degree turn, line up for final, then landing.

 

It works well for me and the traffic flow is orderly -- north to south, then east to west (STAR), then west to east (final) -- no matter what direction the AI is coming from.

 

One important issue I noticed though, and I'll update the docs to reflect this, is don't record waypoint coordinates within FSX when making custom files, FSX does NOT display the coordinates with enough precision (significant digits) for the touchdown points on the runway.  You will need to load up the freeware (donation-ware) Airport Designer X, which can easily provide you with high-precision coordinates for the runway.  Also note you can enter the coordinates in any combination (Airport Design Editor X gives hour decimals, FSX gives hour-minute decimals, LOWI approach charts give hour-minutes-seconds).

 

 

If we get a program the can make the ai taxi, then this program would work without issues. It feels like we ( the flight sim community) are tip toeing around this sleeping monster called fsx. Are we afraid it will bite us? While I'm at it, I will rant some more.

 

Good point.  Making AI taxi is actually easy.  My program can make aI taxi at slow speeds between waypoints, it is easier than landing the AI on a runway.  Also, determining routes from node A (e.g., end of runway) to node B (e.g., gate X) taking into account path metadata (e.g., preferred routes, blocked nodes, etc.) in a network of waypoints (nodes) is also straight forward as there are a lot of well-known and efficient off-the-shelf C++ code to tackle exactly this type of problem (mostly from the world of network routing).  The challenge for me is writing code to read this data from active .bgl files in a versatile manner (e.g., the way the designers of ADEX have done it). 

Edited February 24, 2014 by Clipper Ocean Spray

[ysfsim 22]

 

 

Good point. Making AI taxi is actually easy. My program can make aI taxi at slow speeds between waypoints, it is easier than landing the AI on a runway. Also, determining routes from node A (e.g., end of runway) to node B (e.g., gate X) taking into account path metadata (e.g., preferred routes, blocked nodes, etc.) in a network of waypoints (nodes) is also straight forward as there are a lot of well-known and efficient off-the-shelf C++ code to tackle exactly this type of problem (mostly from the world of network routing). The challenge for me is writing code to read this data from active .bgl files in a versatile manner (e.g., the way the designers of ADEX have done it).

Great. I will run a test to see what happens if the ai is moved to the hold short point after landing without clearance.Is it possible for you to have the ai contact  atc, and ask for taxi to parking?

[Clipper Ocean Spray 189]

Great. I will run a test to see what happens if the ai is moved to the hold short point after landing without clearance.Is it possible for you to have the ai contact  atc, and ask for taxi to parking?

 

I used to have a taxi function in the code.  The program switched some aircraft movement parameters to a "taxi mode" at the end of the rollout to get convincing taxi movements.  Looking back over my archived code, I needed distMOE = 0.00164579 (i.e., a very small MOE) and AIperaircraftcalibration = false, so you would need to add those to your .ini.  However, I'm not sure how well those would work when the AI is flying and landing.

 

My original plan during an autogen recording was to keep recording the AI all the way to the gate.  That way, if the program forced an AI to land without clearance (e.g., no FSX-IAF intercept), at least the AI would taxi to the gate.  However, I thought it was of limited value (multiple AI taxiing to the same one or two gates).  I decided it would be simpler (and just as good from an immersion experience) to simply delete the any problematic AI at the end of the rollout where the user would tend not to even be able to see it happen.  The biggest drawback IMO was if the user stayed at the airport for hours and the gate occupancy went down (remember, new AI traffic will still get injected parked at a gate - so gate occupancy won't go down that much).  How long does the user stay at an airport for hours though?  The longest I stay is about 45 minutes.  Then I taxi for takeoff and the departing AI traffic is taxiing around as usual.

 

-Roland

Edited February 25, 2014 by Clipper Ocean Spray

[FlyTweety 2]

*sigh*

 

OK, you say, it works for you, my opinion is, it don´t.

Maybe i´m to German, expect to much perfection.

 

I tried your procedures, though you use different waypoints, the behaviour is quit the same:

After leaving the STAR and entering the Final the AI switch to the Approach Mode, means e.g., it flys all the way with extended landing gear.

Further is begins to react like complained, 90° turns within a circle of 100 ft, barely able to fly straight but allways meanderings within 10° left/right. By the way, time warp 4x let your control totally collapse in this mode, the AI is turning unpredictable on its way to the next WP.

 

So you might be right when you be convinced "it works" if AIsidstar bring the AI down to the Runway somehow, but to me it is not comparable to "normal" Approaches.

 

Therefore if have again some questions to understand the tool better.

 

1. What is the trigger for a successful handover from the last STAR WP to the call of the FINAL Approach?

    Why did it not work for me to use the STAR to pass the P3d IAF, earn the Runway assignement, continue the STAR down  to the  short final in RWY direction? The only what had to happen is, that at the end of the STAR, (which provides me the smooth 2 min. turns) it need the switch to the FINAL file.

In my opinion all desires are fullfilled: IAF passed, Runway assigned, Runway and first WP of the Final close straight ahead.

2. WHAT is controlling the AI attitude during the Final Mode?

Is it part of the .air file or has AISIDSTAR the full control at this time?

Would it be possible to add triggers into the program to enable us to change the assigned algorythms depending to the waypoint. So that the very fine control e.g. starts at or below 2000 ft AGL?

Like <attitude>NORMAL</attitude>  <attitude>FINE</attitude>

 

3. HOW do you really read the Runway direction?

You told me before, that you first start your sim with customized weather to be sure that all the traffic has the same wind.

I tried it again and again.

Start the sim, switch off OPUS, choose EDDL as the desired Airport, change the time, customize wind to 230° 16 kts change the Aircraft to BOB.

Then i open the flight and move in a position where i can observe the Traffic.

At my networked Notebook i let run FSC 9.5, GPS connected.

I can read in the upper left corner windspeed and direction, 16 kts 230°, like selected.

100% of the Ground Traffic is taxiing to the Runway 23(L)

 

THEN i start the AISIDSTAR Monitor.exe , from there i start AISIDSTAR.

At this time the sim is running about 2 minutes.

The Monitor Window shows all "taxi out" Traffic to RWY 23L, and guess what?

 

Out of 10 incoming Traffic AI´s 3 to 7 will equipped with STAR Files for RWY 05.

At no time the Wind has changed for more then 5 milliseconds, you know, it is fixed in the setting and i can read it all the time.

So, what is wrong?

Also, as mentioned, it don´t turn better, after one hour ore more it seems to turn even worse with less than 30% correct STAR´s.

What can i offer to you that you see that i don´t cheat?

 

 

Regarding to your depression about using your Tool preferred for the "extreme" Airports:

 

What did you expect?

Simple Approaches working satisfiing anyway. OF COURSE we use Aisidstar to crush the limits. circling Approaches, visuals with changing headings, steep Approaches.

And yes, to keep the separation. We really need it not at simple Hillbilly Airports with one Runway, no obstacles arround and one flight per hour..

The fun is to create working Aprroaches under circumstances where FSX/P3d fails.

 

It is no alternative to delete AI before taxiing in, if i don´t want to see that i can allready free the Approach way earlier with FSUIPC.

And if i want it simple, why using decent AI at all?

We could use cubes instead of frameeating Models. Big red cube for Heavys, midrange yellow cube for narrowbodys, small green for Buisness and GA.

FSX does´t care about the shape, it will "fly" anyway related to the config, but with far better FPS...

 

 

You should surrender, the main reason is that we are childish and want to watch traffic as spotters or love to tinker like the guys with their model railway...

[Clipper Ocean Spray 189]

 

 

After leaving the STAR and entering the Final the AI switch to the Approach Mode, means e.g., it flys all the way with extended landing gear.
Further is begins to react like complained, 90° turns within a circle of 100 ft, barely able to fly straight but allways meanderings within 10° left/right.

 

How are you watching the AI do this?  Bringing up a second window to view AI traffic is probably the easiest way, is that what you're doing?  I just need to retrace your steps.  

[FlyTweety 2]

How are you watching the AI do this?  Bringing up a second window to view AI traffic is probably the easiest way, is that what you're doing?  I just need to retrace your steps.  

 

Basically like a Kameleon:

 

Right eye at my Notebook, where FSC runs.

You might see the AI meander left and right about 10°

 

The left eye on the main screen where i follow the AI from anywhere by using large zoom settings.

There i can see that FSC is WYSIWYG (What you see is what you get)

The only thing that is different are the stutters, this is related to the slow hardware of my Laptop and the CPU consuption during recording (single core).

In the Sim the AI really looks like a snake...

 

By the way, in the second half you can see what happens  to me with 4x speed

[Clipper Ocean Spray 189]

 

 

There i can see that FSC is WYSIWIG (What you see is what you get)

 

Is the first part at a 2x sim rate?  I ask because I notice some snaking after turns at a 2x sim rate.  Also, the beginning looks like the AI is in an area where it should be intercepting the FSX-IAF.  Sometimes the AI has to turn abruptly to intercept it because FSX inserts it out of nowhere (you can see it happening in the FSX Traffic Toolbox Map).

 

That being said, I don't think the AI needs to be in "hot" mode during the whole final.  It is easy for me to add a parameter that turns on fine adjustments below a user-specified AGL atltitude, so I'll add that.  I'm also going to add AITurnScalar specifically for final (also controlled by the user-specified AGL altitude).