TANE: Track direction, is it still important?

Christopher824 said:

So let's get past that the track direction matters to almost nobody on this forum.

Click to expand...

That's not correct, and also contradicts the rest of your post.

Christopher824 said:

Forget the track direction markers, which only show you which way the track is facing, if you rotate them they don't change which way the track is facing.

Click to expand...

Correct. But that doesn't mean you can forget them - they can have a significant impact on how certain AI rules behave.

Christopher824 said:

If i build a route where there are 2 different paths to a destination, and one path has track facing the wrong direction, the AI will always choose the route with the track facing the correct direction.

Click to expand...

There's no such thing as "AI". What you have is a set of rules, each one carefully handcrafted for a specific purpose. Which rule are you using? You can't make two routes to the same destination absolutely identical. If you go through a switch then the switch will have a default setting. If you go in different directions then the train facing will affect the operation of the rule.

Christopher824 said:

If I insert some of the XING's or Bridge's with double track, on some Tracks, it will change the track direction of the track you attach it to, for 3 segments. Where before you insert the asset the double track had each track facing opposite directions. After the insert both tracks face the same direction.

Click to expand...

Double track? My guess is that the double track is using track replacement. Any time you replace track it can change direction. You would need to look closely at how both the the track and the crossing are constructed.

Christopher824 said:

What it means to my AI, before I inserted the asset (XING or Bridge), the AI could make it to the destination, after the insert, AI complains, 'Can't find route to destination'.

Click to expand...

If track direction was playing a role here then I would expect the problem would be that the route to the destination would be different, not that it can't find a route. Again, there are too many variables involved to be able to say anything definitive about what might be happening. Patient careful experiment combined with close analysis of the assets involved would be necessary. And that would best be done in a separate thread in the Content Creators section.

SailorDan said:

That's not correct, and also contradicts the rest of your post.


Correct. But that doesn't mean you can forget them - they can have a significant impact on how certain AI rules behave.


There's no such thing as "AI". What you have is a set of rules, each one carefully handcrafted for a specific purpose. Which rule are you using? You can't make two routes to the same destination absolutely identical. If you go through a switch then the switch will have a default setting. If you go in different directions then the train facing will affect the operation of the rule.


Double track? My guess is that the double track is using track replacement. Any time you replace track it can change direction. You would need to look closely at how both the the track and the crossing are constructed.


If track direction was playing a role here then I would expect the problem would be that the route to the destination would be different, not that it can't find a route. Again, there are too many variables involved to be able to say anything definitive about what might be happening. Patient careful experiment combined with close analysis of the assets involved would be necessary. And that would best be done in a separate thread in the Content Creators section.

Click to expand...

#1 Sorry I have contradicted myself
#2 ... Okay, I agree
#3 ... There is such a thing as AI, why would they call it AI if it was not? True, it is a set of rules, but when you say let the AI run from point A to B, then it is AI. It (the program) decides how to get there, so it is AI. So the rule I use is drive to Commodity (Industry Basic or something) and unload or load. Then go somewhere else and do the same thing. It works when I first set it up, but then it breaks after I insert certain assets after it changes the track directions.
#4 Once track is laid, when you attach another asset, the adjacent asset should not be molested, see my previous posts.
#5 and here is the grey area that no one either confirm or deny, just too many variables to actually confirm what I am saying, so I just keep working on it till I can make it do what I want...

Thank You Sir ! I think I tried to report this as a bug, but they wanted better documentation, or something, so I never followed up, just don't have the time, still love this simulation to death...

Christopher824 said:

I think I tried to report this as a bug, but they wanted better documentation.

Click to expand...

That's the point I was making. It doesn't make sense to refer to 'AI' when you are looking at particular issues. So if you reported that 'AI does this or that' then the first question anyone will ask is 'Which AI rule are you using?'. It isn't possible to 'let the AI run from point A to B' - the only thing you can do is to execute a particular AI rule. They are all different, and they all have their own little quirks.

I have now conducted a simple text of the "AI" moving from point A to Point B with a choice of two routes of unequal length using the Driver Commands "Navigate To Trackmark" and "Drive To Trackmark". The two routes had their track laid in opposite directions as shown by the arrows in the screen shot below.



Test 1:
Starting at trackmark "TM West" with the loco facing towards trackmark "TM East" and both switches set for the shortest path.
Driver Command "Drive To TM East" the driver, Andrea, took the straight line and shortest path.

Test 2:
Starting at trackmark "TM West" with the loco facing towards trackmark "TM East" and both switches set for the shortest path.
Driver Command "Navigate To TM East" the driver, Andrea, took the straight line and shortest path.

Test 3:
Starting at trackmark "TM West" with the loco facing towards trackmark "TM East" and both switches set for the longest path.
Driver Command "Drive To TM East" the driver, Andrea, took the straight line and shortest path resetting the switches as she went.

Test 4:
Starting at trackmark "TM West" with the loco facing towards trackmark "TM East" and both switches set for the longest path.
Driver Command "Navigate To TM East" the driver, Andrea, took the straight line and shortest path resetting the switches as she went.

Test 5:
Starting at trackmark "TM East" with the loco facing towards trackmark "TM West" and both switches set for the shortest path.
Driver Command "Drive To TM West" the driver, Andrea, took the straight line and shortest path.

Test 6:
Starting at trackmark "TM East" with the loco facing towards trackmark "TM West" and both switches set for the shortest path.
Driver Command "Navigate To TM West" the driver, Andrea, took the straight line and shortest path.

Test 7:
Starting at trackmark "TM East" with the loco facing towards trackmark "TM West" and both switches set for the longest path.
Driver Command "Drive To TM West" the driver, Andrea, took the straight line and shortest path resetting the switches as she went.

Test 8:
Starting at trackmark "TM East" with the loco facing towards trackmark "TM West" and both switches set for the longest path.
Driver Command "Navigate To TM West" the driver, Andrea, took the straight line and shortest path resetting the switches as she went.

The only way I could get the driver to take the longer path, regardless of which end she started from, was to precede the "Drive To" or "Navigate To" command with a "Drive Via Trackmark TM Loop" or "Navigate Via Trackmark TM Loop" command.

I could try and set up a test where both track branches are of identical length but this would be difficult to do and I am not sure that it would prove anything of practical value.

Some here will never change there way of thinking on this,And that's ok,Not a problem to anyone.

I am one of those..I believe its a myth and you won't ever change it.
I have played Trainz since 2008 and never bothered with the direction theory....Any trackside object's or train's that faced the wrong way were just a click away on the mouse to alter this.
Everything worked as always.

Some of my track have different directions between splines on the same line.doesn't make a train run up to it and turn back half way down the line..No

When tracks are made they have to have a begining and a end ..a start and a finish ..and same when laid..But it doesn't make a difference to the running of trainz

First, let me restate my last post where I said "The myth that track can be laid in the wrong direction is exactly that - a myth."
- restated - The myth that track can be laid in the wrong direction, thus causing AI to malfunction, is exactly that - a myth"

pware said:

I have now conducted a simple text of the "AI" moving from point A to Point B with a choice of two routes of unequal length using the Driver Commands "Navigate To Trackmark" and "Drive To Trackmark". The two routes had their track laid in opposite directions as shown by the arrows in the screen shot below. . . . . .

Click to expand...

Thank you for posting your test. The results are exactly what should be expected. AI is artificial intelligence. It is designed and expected to take the shortest route to get where it has been told to go. If you want the train to take the siding, you have to tell it to do so. It is "artificial" intelligence, not mind reading intelligence.