FSX-MS Ai Controller (Sid, Jet Route, Star And Final Approach) Beta

[Clipper Ocean Spray 189]

Thanks for the reply. I have enabled both options as suggested but I still find AI going all over the place to intercept IAF. From AI Monitor, the flight status for AI with such behaviour is shown as "Intercept IAF" and controller status is FSX instead of AI Controller. :(

 

Is this happening early on, before an auto-gen final approach file is recorded?  If so, the AI will always revert to FSX controller because there is no final approach file to read.  At airports with lower traffic levels, it may take some time before autogen finals are recorded for all of the typical landing runways.  However, once they're recorded, they're available to immediate re-use the next time the program is run.

[jabloomf1230 2,445]

I didn't want to leave anyone with the impression that RC4 doesn't work properly with P3d. It worked fine for me up until the 2.3 update. Now, for some reason sometimes I can't get it to respond, but the problem occurs randomly. I may need to reinstall RC4.

[Clipper Ocean Spray 189]

 

 

I didn't want to leave anyone with the impression that RC4 doesn't work properly with P3d. It worked fine for me up until the 2.3 update. Now, for some reason sometimes I can't get it to respond, but the problem occurs randomly. I may need to reinstall RC4.

 

Understood, I'm glad you and others bring this stuff up...I'm looking to controlling taxi operations and it is good to read these comments and be reminded that voice chatter (specifically for AI on the ground) is a feature that I probably need to maintain if at all possible.

 

-Roland

[gfd 15]

Thanks Roland,

 

Re Strange character set: Language is Canada English. Keyboard is US

[gfd 15]

Hi Roland,

 

I would like to get the most out of AICull. Let's say I set up a flight from CYVR to CYYC. Assume the ini file only has procedures for these airports, and is therefore monitoring no others. Assume that the necessary ini entries and data file are present for enroute control. Assume AICull is set to delete ground traffic only.

 

Now having ground traffic for the above airports (only) makes sense. But what is the effect on airborne traffic? Will the sky be populated only with aircraft that have departed from, or are arriving to, the 2 airports?

 

What I am looking for, is a scenario where AI ground traffic is limited to arrival and departure airports, to lessen the load on the sim; but with realistic levels of airborne traffic.

 

Separate but related: if I were to use the world-wide ini file I created, would it have the practical effect of crippling AICull; since all airports would be monitored?

 

Thanks not only for the great program, but for taking the time to support your creation,

 

Graham

[Clipper Ocean Spray 189]

I would like to get the most out of AICull. Let's say I set up a flight from CYVR to CYYC. Assume the ini file only has procedures for these airports, and is therefore monitoring no others. Assume that the necessary ini entries and data file are present for enroute control. Assume AICull is set to delete ground traffic only.

 

Now having ground traffic for the above airports (only) makes sense. But what is the effect on airborne traffic? Will the sky be populated only with aircraft that have departed from, or are arriving to, the 2 airports?

 

What I am looking for, is a scenario where AI ground traffic is limited to arrival and departure airports, to lessen the load on the sim; but with realistic levels of airborne traffic.

 

 

 

Setting the AICull options to delete ground traffic will give you the scenario you're looking for, AI ground traffic limited to monitored airports (in your case, the arrival and departure airports), but airborne traffic unaffected.

 

 

 

Separate but related: if I were to use the world-wide ini file I created, would it have the practical effect of crippling AICull; since all airports would be monitored?

 

 

 

Good question, if you're using a world-wide ini file (e.g., hundreds of monitored airports), then that would indeed work against the aims of the AICull option in AIController.  Fortunately, there's a stand-alone version of AICull in the library (AICullv11.zip) so that you can enter a separate list of ICAO(s), such as your actual departure, destination, and perhaps an alternate airport.  That way, you don't have to fiddle around with your world-wide AIController.ini (you also don't have to use AIController at all).  

 

-Roland

[netshadoe 144]

'netshadoe', on 02 Sept 2014 - 12:38 PM, said:

 

On another note, is there a cap on how many AI the program can control? I've had a few issues (mainly with jet routed AI) where AIController is controlling the AI (according to the monitor), yet the AI does not respond at all to control commands. (I can see them going on their merry way on Flight Sim Commander). I'll get a mix...some are controlled, some fly under FSX control.

 

No cap, but I'll look into this, is there an airport and situation where you usually see this? Are the AI not being controlled far out from the user's current position or close in?

 

Good question.  I'm pretty sure it's when the AI spawns entering the reality bubble.  I'll do some testing to see if it only occurs to these flights.  I never noticed it with any flights that originate within that bubble (either spawned inside the bubble or taking off from an airport within the bubble)

 

 

I've been testing the jetroute function of this program more than others (since you do have many helping you out on the SIDS/STARS aspect of it.    I've noticed a few things that don't work or can be improved:

 

- The flight levels still don't work properly.  I've tested this while both sitting and AI spotting, and during flight.  I was flying from CYOW to KLAX last night (westbound at FL340) and had many "encounters" with AI flying in the opposite direction at my level when they should be 1000 above or below.

 

- Going back on my suggestion about separation, could the program also separate AI jet traffic from each other too?  Again, on my flight last night, I had 3 AI traffic at the same level one right after the other (within 3-4 plane lengths from each other) like they were flying in formation!!  

If AI inside jetroutes (not SIDS/STARS...keep those separate) enters another AI's 5 mile bubble, the offending AI should be moved 2000 feet up or down (next appropriate flight level) to avoid conflict.

 

- I'm still testing this, but I'm still having problems with AI finding a route over oceanic airways in some instances.  I can't pinpoint as to why some work and most don't.  Any suggestions on what I can test to help you out?

 

 

All in all, you are ground breaking with this program Roland!!  It is neat to be flying along an airway and have a 747 barreling at you a couple of thousand feet above you on the same airway!!  And the SID/STAR function is coming along nicely and is also unique in allowing us to follow other AI in the "pattern".

 

Let me know if you have any suggestions or want anything during my testing that can help you.

[Driftingisfun106 0]

You don't have to buy anything, just like you don't have to use this fantastic application if you don't want to.

 

Unless this app takes over every stage of flight including ground operations (which I don't see happening as it's far too complex of a thing to do, it took Microsoft a decade or more to get to this point) then there is no way it can stop traffic from stopping a long way behind you on the ground. That is a function of the AFD file or AFCAD as it used to be called. The person who built the AFD file didn't put enough nodes on the taxiways, so in your case if a plane really did stop 1000 meters away (yes I know you are exaggerating massively ) that was because the nodes on that taxiway were 1000m apart.

 

 

I haven't experienced this, but I'll keep an eye out for it.  Let me know if you find out any new information.  Unfortunately, although AIController will not make any permanent changes to FSX, it is possible that programs interfacing through simconnect to FSX can crash FSX.  I ran into this problem with an earlier version (now deprecated) that assigned temporary flight plans to the AI.  

 

 

 

 

I don't control taxiing now, but that's my ultimate end goal.  There's many interesting taxi parameters that I want to give the user control over and your suggestion is a good example:  the clearance area required by a taxiing AI.

 

-Roland

Man I'm so late LOL been a while AVSIM anyhow I got what you guys are saying about nodes and such but I just hate that the ACES did it on purpose because in fs2004 people were complaining about the ai not stopping and crashing into you or something. But anyhow, I understand what you are saying. I may also get ADE just to get rid of nodes or what not haha. 

[gfd 15]

Setting the AICull options to delete ground traffic will give you the scenario you're looking for, AI ground traffic limited to monitored airports (in your case, the arrival and departure airports), but airborne traffic unaffected.

 

 

 

 

 

Good question, if you're using a world-wide ini file (e.g., hundreds of monitored airports), then that would indeed work against the aims of the AICull option in AIController.  Fortunately, there's a stand-alone version of AICull in the library (AICullv11.zip) so that you can enter a separate list of ICAO(s), such as your actual departure, destination, and perhaps an alternate airport.  That way, you don't have to fiddle around with your world-wide AIController.ini (you also don't have to use AIController at all).  

 

-Roland

 

Thanks Roland,

 

I think I'll use the full version of AI Cull with AI Controller. That would be the best of both worlds.

 

Do have another question (please advise when question quota used up ^_^)

 

I would like to know how I might get an idea of what aircraft should be in the control of AI Controller as opposed to FSX. Are there criteria I can use to determine this?

 

Thanks again,

 

Graham

[gfd 15]

Bumping my last post. Although AI Controller states FSX controlling, etc, etc, I am using P3D V2.3.

 

Thanks

Graham

[Clipper Ocean Spray 189]

The flight levels still don't work properly.  I've tested this while both sitting and AI spotting, and during flight.  I was flying from CYOW to KLAX last night (westbound at FL340) and had many "encounters" with AI flying in the opposite direction at my level when they should be 1000 above or below.

 

 

I'll look into it, maybe I should just key off the AI's current heading.  Right now, I take the average heading of the computed jetroute (sometimes leading to counter-intuitive results for long jetroutes), but a simpler algorithm might actually work better.

 

BTW, did the AI flying in the opposite direction at your flight level get close to you and then ascend or descend to avoid you?   Just wondering what people are seeing.  The flight-level collision avoidance feature is hard for me to test and debug.  I'm curious whether I should increase the default flight-level avoidance radius to 5 nm or even greater.

 

 

 

If AI inside jetroutes (not SIDS/STARS...keep those separate) enters another AI's 5 mile bubble, the offending AI should be moved 2000 feet up or down (next appropriate flight level) to avoid conflict.

 

 

Yes, now that you put it that way, that might a relatively easy fix (just change the flight levels).  I wanted to avoid the complexity of another distance-based separation routine similar to what I used for separation AI on STARs, so yeah, thanks for the idea.

 

 

I'm still testing this, but I'm still having problems with AI finding a route over oceanic airways in some instances.  I can't pinpoint as to why some work and most don't.  Any suggestions on what I can test to help you out?

 

 

Maybe you could try tweaking the favored and disfavored jetroute prefixes and "weights".  It could also just be the incompleteness of the Northern Atlantic and Pacific tracks I added (e.g., it wouldn't hurt if there were more mid-Atlantic tracks).  Do you use the skyvector.com site?  I like it a lot.

 

 

I would like to know how I might get an idea of what aircraft should be in the control of AI Controller as opposed to FSX. Are there criteria I can use to determine this?

 

 

The main criteria are whether the airport is monitored, if so, the broad parameters to next look at (readme, pp. 4-14) are AIMaxSpawnDistanceFromSTAR, AIMaxSpawnDistanceFromSID, AIClimbOutHeight, AIMaxSpawnDistanceFromAirport parameters.   There's also more specific (scenario-based) parameters, like AIGoAroundLeaveArea, AIFinalApproachUseFile, AIShortFinalRelease, etc.... .  The flight stages typically controlled by AIController:  jetroutes, STAR, terminal approach, holding, final approach (if custom or autogen file found), and SID.  The flight stages typically controlled by FSX:  IAF intercept (unless optionally bypassed), final approach (if NO custom or autogen file found), go-arounds, and the initial climb-out on departure.   

 

 

Bumping my last post. Although AI Controller states FSX controlling, etc, etc, I am using P3D V2.3.

 

 

Yes, when using with P3D, "FSX controlling" means P3D controlling.

 

-Roland 

[netshadoe 144]

 

 

I'll look into it, maybe I should just key off the AI's current heading. Right now, I take the average heading of the computed jetroute (sometimes leading to counter-intuitive results for long jetroutes), but a simpler algorithm might actually work better.

 

That would probably be your best bet and most likely would cause you less headaches!!

 

 

 

BTW, did the AI flying in the opposite direction at your flight level get close to you and then ascend or descend to avoid you? They should have, just wondering what people are seeing.

 

No.  Not at all.  The three AI in a row that I mentioned were close enough to me that I could tell is was United, Southwest, United!!    I would say most AI passed within a 1/4 mile or closer on the jet route. 

 

Should the option UserAIseparationdistance=0.0 be on for this to work, or is that just for the SID/STAR part of the program? 

 

In my case, it was as you see it (off).

 

 

 

This is a hard feature for me to test and debug by myself. I'm curious whether I should increase the default (flight level) avoidance radius to 5 nm or even greater.

 

 

 

Yes, now that you put it that way, that might a relatively easy fix (just change the flight levels). I wanted to avoid the complexity of another distance-based separation routine similar to what I used for separation AI on STARs, so yeah, thanks for the idea.

 

Again, keeping the jetroute and SIDS/STAR parts separate, maybe conflicting AI on the jetroutes within 15NM of each other should be commanded to climb or descend.  One AI should be given priority (maybe sort them as eastbound traffic is priority for the sake of simplicity), the other AI (the westbound) should be the conflicting AI that has to move out of the way.  I use 15nm because a 2000 ft level change can take about 8-10 miles at 1000ft/min or so, thus giving the conflicting AI enough time to move realistically.

 

 

Maybe you could try tweaking the favored and disfavored jetroute prefixes and "weights". It could also just be the incompleteness of the Northern Atlantic and Pacific tracks I added (e.g., it wouldn't hurt if there were more mid-Atlantic tracks). Do you use the skyvector.com site? I like it a lot.

 

I'll give that a shot.

 

Just for fun, I'll also try fooling around with NATS and the like in the oceanic waypoint database to see if I can get some consistency.

 

I've glanced at skyvector, but haven't used it extensively.  I use Flight Sim Commander to show the jet routes.  Looking at it now, I think I'll use it to experiment with waypoints, jetroute prefixes and weights.

[ysfsim 22]

Roland, could you make the aiconv be able to extract approaches as well as sids and stars? This comes in handy in places like LOWW where the transitions from the stars to the approach are winding paths...I know you will need runway data to complete an approach file but could you leave that to be manually added so that we can get the waypoints of the approach extracted? thx

[Clipper Ocean Spray 189]

No.  Not at all.  The three AI in a row that I mentioned were close enough to me that I could tell is was United, Southwest, United!!    I would say most AI passed within a 1/4 mile or closer on the jet route. 

 

 

Thanks, looks like I need to increase the default avoidance radius then.

 

 

Should the option UserAIseparationdistance=0.0 be on for this to work, or is that just for the SID/STAR part of the program? 

 

 

The UserAIseparationdistance is just for the SID/STAR part of the program, correct.  The collision avoidance at flight levels (jet routes) is independent and default on.

 

 

 

Roland, could you make the aiconv be able to extract approaches as well as sids and stars? This comes in handy in places like LOWW where the transitions from the stars to the approach are winding paths...I know you will need runway data to complete an approach file but could you leave that to be manually added so that we can get the waypoints of the approach extracted? thx

 

 

Definitely, you've hit upon my plan!  Others have suggested this to me as well.  The change in my viewpoint is I recently realized (after researcing into taxi operations) that the runway waypoints can be easily and automatically discovered, so there's no need to manually add runway data.  The approach conversion can be fully automatic, like it is for SID/STARs.  Matias' conversion utility will only need to be modified to reference the new aiconv.exe file.

 

This also means better (generic) autogens can be constructed when there is no approach navdata to convert and the user has not written a custom approach.  First, the autogen will no longer have to be "recorded."  Second, the user will be able to have more parametric control over the autogens (e.g., final length, intercept alt, turn angle onto final).  

 

-Roland

[ysfsim 22]

Definitely, you've hit upon my plan! Others have suggested this to me as well. The change in my viewpoint is I recently realized (after researcing into taxi operations) that the runway waypoints can be easily and automatically discovered, so there's no need to manually add runway data. The approach conversion can be fully automatic, like it is for SID/STARs. Matias' conversion utility will only need to be modified to reference the new aiconv.exe file.

 

 

 

This also means better (generic) autogens can be constructed when there is no approach navdata to convert and the user has not written a custom approach. First, the autogen will no longer have to be "recorded." Second, the user will be able to have more parametric control over the autogens (e.g., final length, intercept alt, turn angle onto final).

Oh nice! btw, how many approach files can be active for one runway? If an ai leaves STAR alpha and goes to bravo transition for approach charlie on runway 23, can another ai leave STAR zulu, go to golf transition for approach charlie to runway 23 ( the same approach but it has two different transitions, therefore two files, just like the SIDS and STARS)

 

 

I'm seeing ai on final under ai controller's control perform go arounds instead of landing due to ai on the ground not holding short for landing ai. How to fix that? I thought the incursion line was to freeze ai that are ready to depart if one is on final?