How does the AI work in TRS22?

[MartindeGroot]

Is there a good, comprehensive documentation about the inner workings of the AI in the current version of Trainz, TRS22?
I want to know why a train is waiting for a signal, when I can't find a good reason.
Also,, when an AI train has control over a junction, I want to know how far ahead these junctions are controlled )(and blocking a manual train from changing them) and when the control over a junction is released?
I am at the moment trying out basic things like having two trains on the same trrack, where the first train waits at a side-track station and want to see if the second train will pass the first train on the main track. So far this works, but it would really help if there is a description of how AI trains operate with regard to changing junctions, and waiting for red signals etc. Some older link which I found in an older post is no longer working.
Any help will be appreciated.

 

[ColPrice2002]

From my experience with the Edinburgh to London route (ECML - cloned & edited) it can be tens of miles ahead - patience!

 

[pware]

it can be tens of miles ahead

I forget the exact figure but that sounds about right. The AI will "look ahead" a looooong way to find the next signal or switch. If, for example, the next signal is (say) 30kms ahead and it is set at STOP then the train will slow to about half speed for that entire trip. Where there are large distances between signals I always place an invisible signal or two a few hundred metres ahead of that next signal. That way the AI train will see the next signal at PROCEED (if 2 invisible signals are used) or CAUTION (if only one is used) and it will move at the set line speed.

Likewise an AI train will grab control of any switch that far ahead. There are ways around that but they involve using session rules and are more complex.

 

[ColPrice2002]

It also depends on the possibilities of route markers, other trains and it will also depend on the possible alternative routes..