Railworks America

Direct an AI engine to go to a destination on the other end of a yard. Assume that all but one track through the yard is occupied. Invariably the program selects an occipied track and crashes into whatever is occupying that track. In short an AI engine does not detect an occupied track except when occupied by another engine. Is this an acknowledged problem?

A workaround is to direct the AI (use timetable) to go By Way Of (Waypoint) the clear track to get to the other end of a yard.

I'm not sure if I completely understand what you are trying to say, but yes, you can use the timetable.

It's the same thing with crossing gates. If your train is long enough, the gates will go up before the whole train passes by because the cars aren't acknowledged. I'm not sure if DTG intended this or not, but as far as I know, it probably more work than it's worth to fix this.

StrategyFirst wrote:I'm not sure if I completely understand what you are trying to say, but yes, you can use the timetable.

It's the same thing with crossing gates. If your train is long enough, the gates will go up before the whole train passes by because the cars aren't acknowledged. I'm not sure if DTG intended this or not, but as far as I know, it probably more work than it's worth to fix this.

You are correct, it happened on Donner Pass and all routes. Gates go up and after a few seconds they go down.

Accurate wrote:Direct an AI engine to go to a destination on the other end of a yard. Assume that all but one track through the yard is occupied. Invariably the program selects an occipied track and crashes into whatever is occupying that track. In short an AI engine does not detect an occupied track except when occupied by another engine. Is this an acknowledged problem?

A workaround is to direct the AI (use timetable) to go By Way Of (Waypoint) the clear track to get to the other end of a yard.

What you've described does happen at times, yes, but the fix you've also described is quite well known, so I'm not sure what your point is.

I apologize for being unaware of the existing fix. I just tossed that in there encase someone might benefit, not realizing it was an annoying duplicate.

An AI engine not recognizing a car/wagon has been around for quite awhile and I may have missed any fix or workaround for it.

Another oldie is not being able to couple to a car more than once. There is a workaround for that somewhere. Wonder if it was ever addressed.

Accurate wrote:Another oldie is not being able to couple to a car more than once. There is a workaround for that somewhere. Wonder if it was ever addressed.

Either I don't understand what you mean, or it's just you.

In a typical rail operation a switcher might pick a car from the yard, drop it at a customer's siding for loading. Later the switcher will pick the car take it to the yard and add it to a building long haul consist.

The old issue was that once a couple to a car and uncouple took place you could no longer couple to that car. However, if you performed the second coupling, with the coupler over a track marker, it would work.

I was wondering if that remains the case.

I am sorry for having been so confusing and imprecise. I thought most would have used that maneuver to make up consists while collecting laden customers cars.

Accurate wrote:In a typical rail operation a switcher might pick a car from the yard, drop it at a customer's siding for loading. Later the switcher will pick the car take it to the yard and add it to a building long haul consist.

The old issue was that once a couple to a car and uncouple took place you could no longer couple to that car. However, if you performed the second coupling, with the coupler over a track marker, it would work.

I was wondering if that remains the case.

I am sorry for having been so confusing and imprecise. I thought most would have used that maneuver to make up consists while collecting laden customers cars.

I don't recall ever having this problem. Believe me, I experiment a lot! Could you make a video of what you mean?

Accurate: What you are saying is the way it still is, nothing along those lines have changed. So you have to work around with the waypoints, and you can not plan multiple uncouple/couples with out grief. It ain't gonna happen. There have been no real changes to anything AI. I am assuming that will all come with the UE4 version in a year or 2.

artimrj wrote:Accurate: What you are saying is the way it still is, nothing along those lines have changed. So you have to work around with the waypoints, and you can not plan multiple uncouple/couples with out grief. It ain't gonna happen. There have been no real changes to anything AI. I am assuming that will all come with the UE4 version in a year or 2.

Now Bob...
You know as a former Marine, not to a$$/u/me anything.

A person can't spell the short name of a Donkey around here?

Be nice we have lots of children around here. The censor is tough.

Accurate wrote:In a typical rail operation a switcher might pick a car from the yard, drop it at a customer's siding for loading. Later the switcher will pick the car take it to the yard and add it to a building long haul consist.

The old issue was that once a couple to a car and uncouple took place you could no longer couple to that car. However, if you performed the second coupling, with the coupler over a track marker, it would work.

I was wondering if that remains the case.

I am sorry for having been so confusing and imprecise. I thought most would have used that maneuver to make up consists while collecting laden customers cars.

==============================================

Based on my experience, all AI switching operations are "accidental", meaning that they were not foreseen by the programmers. Given that, the following AI switching operations are usually possible:

1) An AI engine can couple to any piece of rolling stock that has not moved or been moved since the scenario has started running, provided that the rolling stock to be picked up is located on a track marker. That has been and still is an inflexible rule. Believe me, I've learned this the hard way.

2) An AI engine can drop off one or more pieces of rolling stock, provided that the drop occurs on a track marker. Once dropped off by an AI engine, rolling stock cannot be picked up again by the AI engine that dropped it or by any other AI engine. It can, however, be picked up by the player engine (sometimes).

3) An AI engine can pick up rolling stock from more than one track marker (but see rule 1 above) and can make multiple drop-offs on multiple track markers. However, the AI engine must be given very precise routing instructions!

4) An exception to rule 3 is that an AI engine can only pick up one string of rolling stock from any one track marker. However, multiple pickups can sometimes be made from multiple track markers located along the same track.

5) This rule trumps all others: all AI switching is subject to the whims of what is often referred to as "the dispatcher".

I've found that AI switching that works in one scenario may not work in another scenario located in the same place and that AI switching that works just as I planned it may well fail to work after a subsequent editing session, especially if that session involves adding additional AI traffic. In other words, leave AI switching maneuvers until all other AI and player activities have been thoroughly tested. As we all know, AI stands for Artificial Intelligence; in this simulator, a more proper term would be DL, standing for "Dumb Luck". Happy scenario writing to all with the nerve to try and survive it!

A good summation Tom. Positioning the coupler(s) over a marker is the only way to grab a car for a second or subsequent time. As you said the same marker may only work once. It is a pain to do, but the only way, if a switcher places a car in a yard for later inclusion in a long haul consist.

If this anomaly did not exist then scenarios of an engine going to certain companies to pickup loaded or empty cars and then making up a new consist with those cars and others. Since it does, the yard would be full of markers spaced at many different places to not only accomplish the task but to accommodate various lengths of cars. Ugh.

Accurate wrote: ... the yard would be full of markers spaced at many different places to not only accomplish the task but to accommodate various lengths of cars. ...

Again, in my experience, the best way to lay out a yard at the route building level is to place a short marker as close to the entrance or exit switch of the first siding in each section of the yard (e.g.: Martindale North 1, Martindale South 1, etc.) to let scenario authors understand the scenario maker's intentions. This allows scenario authors to place any and all additional markers needed for a particular scenario. It is often helpful to place more than one track marker on the same siding at the scenario editing level, which is exactly the situation that you seem to find offensive. May I ask whether you write scenarios?
In any case, end-to-end siding markers are the bane of a scenario author's existence.

I usually make changes to a route to improve its appearance - my personal view of the world. Then I try various scenarios with a goal of running a few trains at once if the simulated railroad is appropriate for that type of operation. Otherwise, I may just tag along for the ride (CTRL_CLICK on engine) and enjoy the scenery. If there is a yard and there are lots of markers for various coupling events then it becomes a bit messy. Not impossible but neither is it prototypical, which is a drawback for me. It is all personal preferences.

How real does a route look and then attempt real-world operations. It is all for personal use. No "publishing".