AI train locks switch, manual train boxed in

I'm looking for help with this situation.


While a manually operated train was waiting in the headshunt track 5, an AI train has arrived on track 2. The AI train is scheduled to go to a trackmark in track 5, which is occupied. So correctly, the exit signal from track 2 shows red.


I want to throw the switch (turnout) marked with the pink circle, but the turnout lever "is controlled by an AI train" and I cannot change it. Therefore the Nohab train cannot free up track 5 by proceeding to track 1.


(Track 3 is never used. Nor is the straight stub at the end of track 1.)


Checkmate! End of game.


There is something clever I should do here, with intermediate trackmarks or signals, but I'm at a loss. Any suggestions would be welcome!



AI-train-locks-up-headshunt by Rob Kievit, on Flickr

I've seen that.

What kind of signals are these.

I see two absolute signals on tracks 1 and 2.

What kind of signal is that in the distance for the manually driven train.

You might want to try a shunting signal, or an absolute signal on the track 3, and remove the signal at the top part of "No problem".

How can you visually tell an absolute signal, from a permissive shunting signal ?

Just because a signal has a number plate, or does not have a number plate, does not fully insure that it is permissive, or an absolute signal, I'm sure there are exceptions

How does an absolute signal differ, from a permissive shunting signal, and what lines in a config file (or script) can be altered make an absolute signal, behave like a permissive signal ?

I usually use BNSF50 Invisible Signal, or Invisible Signal Thingy, as a lot of prototype tracks are not signaled by a visual signal head

MP242 said:

How can you visually tell an absolute signal, from a permissive shunting signal ?

Just because a signal has a number plate, or does not have a number plate, does not fully insure that it is permissive, or an absolute signal, I'm sure there are exceptions

How does an absolute signal differ, from a permissive shunting signal, and what lines in a config file (or script) can be altered make an absolute signal, behave like a permissive signal ?

I usually use BNSF50 Invisible Signal, or Invisible Signal Thingy, as a lot of prototype tracks are not signaled by a visual signal head

Click to expand...

The rule is... Permissive/home signals have a name board while absolute signals do not. Type 05 versus Type 04 in N3V/JR parlance.

If someone doesn't follow the rules, then it's another issue altogether and that asset is bogus.

I commonly use PRR PLS02L Style B Gantry - seniorchief, and PRR Dwarf Signal - gfisher, how do I tell if these are absolute, or permissive signals, as they have no number plates (although they actually may be permissive signals) ?
As the config files give absolutely no clue whatsoever what the naming means, ones is forced to guess what it is used for.
How would you test a signal to ascertain whether it is absolute, or permissive ?

MP242 said:

I commonly use PRR PLS02L Style B Gantry - seniorchief, and PRR Dwarf Signal - gfisher, how do I tell if these are absolute, or permissive signals, as they have no number plates (although they actually may be permissive signals) ?
As the config files give absolutely no clue whatsoever what the naming means, ones is forced to guess what it is used for.
How would you test a signal to ascertain whether it is absolute, or permissive ?

Click to expand...

If the signal is permissive, aka distant signal, the signal will change aspects, to reflect the condition ahead, but the AI will pass them. Example: There's a red signal shown on a permissive signal. This means the upcoming signal is red. If the permissive signal is yellow-flashing, it means the next signal is going to be a solid yellow, which then in turn means the next one is red, and then finally the absolute signal is red.

See how this cascades down.

If you are using the incorrect signals, then you are going to have problems. The other problem is, and I forgot to mention this before, if they are older signals, they use the older unscripted, but look like the real thing but are not, operation. This means they do their own thing as simple signals, and do not follow any rules.

N3V like Auran later implemented, the NORAC operating rules for US signaling. Other systems work differently, and many of these signals are not scripted like the new ones are.

Have the computer controlled train drive to a trackmark place before the signal that is stopping it. That will leave the switch free until the player train has passed. Once the computer controlled train gets a green signal it will then lock the switch and proceed on its way. The rule is simple, a computer controlled train with a driver command to drive to any destination beyond a switch will lock the first switch regardless of how far away it is when that command is executed. So delay the execution of the command to as late as possible.

William

Possibly just before the turnout, you could add another secondary invisible signal, in front of the signal where the AI train is shown waiting, thereby unlocking the turnout ?

As track 1 is shiny rails, I presume it is the mainline track, and aligning all turnout default positions, in surveyor, lined toward the mainline track, might help simplify things

If all else fails command the AI driver to 'stop' (control-right click) and they will release the switch/point/turnout so the other train can proceed. Once that happens you can command the AI driver to 'continue schedule'.

Thank you for the helpful hints, folks.


@KotangaGirl's workaround - temporarily remove AI control from the train on track 2 and change the junction - works OK. That at least is a workable fallback option. Now let's try and prevent the AI from locking the junction beyond the signal.


I tried removing all the signals between the exit signal of track 2 and the headshunt, track 5. The AI still grabs control of the crucial junction with the pink circle.


I inserted an invisible junction immediately after the exit signal of track 2, to no avail, as the AI grabs control of both that junction and the pink-circled one.


Next I ripped out the invisible junction and inserted a trackmark between the track 2 exit signal and the junction that I want to avoid being taken under AI control. That's recommended in this Trainzportal blog post http://www.trainzportal.com/blog/view/using-the-trainz-ai-in-sessions-part-2 .


Still the AI train halts before the signal, correctly, but the turnout lever remains under control of the AI train. It's maddening. I'll keep experimenting.


Maybe the cause of the problem is that everything is too close together, and the AI train is moving slowly, so everything is always 'within braking distance' - an important factor in AI behaviour, I gather from the abovementioned blog post.


@JCitron, the red semaphore signals, by the way, belong to the Netherlands' 1937 system <Kuid2:452956:100511:1>. They are scripted, and the rule is called SignalScriptManager <kuid2:75834:60000:1>.

Thank you Rob for that info. I'll check out those signals.

Could you try a track mark on track 1 with directions for the AI driver located a 5 to take that line?

Hint: put a track mark at 1, 2, and 3, and of course have the driver take the appropriate one. This sadly is one of those hand-holding situations.

Thank you John. I guess that in the course of the discussion it got overlooked that the train on track 5 is the player-operated one.

Your solution would presumably be perfect if that was also an AI train. Then the AI would keep trying to move either the train on track 5 or the one on track 2. In order to clear track 5, the AI would throw the junction towards track 1 with the trackmark you proposed, and allow the Nohab train to 'escape'.

However... the AI train in track 2 now simply sits and waits for track 5 to be vacated by the player-run train, which is impossible because the junction towards track 1 remains locked in the direction of track 2, by the AI train there.

It's an interesting puzzle. On the one hand, I want the train on track 2 to wait before the red signal, which it does, because looking ahead it sees that way beyond, the track is occupied by another train. But on the other hand, I don't want that same train to look ahead to the nearest junction and lock it.

I tried all the solutions mentioned earlier in the thread, but so far, no success.

I've had similar trouble, and it also depends on how you want to run this, presumably as a session. What you can do, and it's not perfect is set up a trigger so your train takes control of that junction, and when using 'Set junction' command, lock it to prevent AI control, with another trigger after you have passed the junction to release control. Timing can be critical if both trains are there at the same time, easier if your train arrives slightly earlier.

RobKi said:

Thank you John. I guess that in the course of the discussion it got overlooked that the train on track 5 is the player-operated one.

Your solution would presumably be perfect if that was also an AI train. Then the AI would keep trying to move either the train on track 5 or the one on track 2. In order to clear track 5, the AI would throw the junction towards track 1 with the trackmark you proposed, and allow the Nohab train to 'escape'.

However... the AI train in track 2 now simply sits and waits for track 5 to be vacated by the player-run train, which is impossible because the junction towards track 1 remains locked in the direction of track 2, by the AI train there.

It's an interesting puzzle. On the one hand, I want the train on track 2 to wait before the red signal, which it does, because looking ahead it sees that way beyond, the track is occupied by another train. But on the other hand, I don't want that same train to look ahead to the nearest junction and lock it.

I tried all the solutions mentioned earlier in the thread, but so far, no success.

Click to expand...

I've actually had the same situation on a tram route I operate extensively. Is there a double-track near your train? If so, then set the switch as default for that direction. What that will do is get the AI driver out of your way automatically then you can proceed. The alternative to this is to setup path control and triggers.

There are other things to try, but I need to look at how I approached them on my route.