I know that "AI" is just a buzzword that is used on the forums to describe a feature. But the operation, or lack thereof, of AI is not really an artificial intelligence problem. It is rather a topology challenge. Now Trainz allows users to create layouts with hundreds of baseboards, and thousands of trackmarks and junctions, most of which are not even loaded at any given time. I think the solution, if any, lies beyond incremental coding changes. I suggest it is possibly a product unto itself.
Disappointed with TRS19 again.....Help pls
- Thread starter marv999
- Start date
AI can only ever be as smart as those who program it - including our users.
Most issues occur because people don't use the right command at the right time, or place track marks strategically, or fail to fully consider the behaviour options perceptible to the AI driver within the route & sessions created by Trainz users.
I never blame AI for failures on my watch. It is always something I failed to do. Correct those things - and Drivers behave themselves assiduously, every time.
TRS19 is easily the best performing, most attractive, and photo-realistic version of Trainz so far to date.
Let's get these minor issues/ bugs dealt with quickly so that we can all enjoy this railroad simulator supremacy together.
The Trainz AI can certainly be made to work with few or even no issues - even with quite complex journeys over complex junctions and many train interactions. As you say, one necessary (but insufficient) condition is to write AI programs that have "the right command at the right time". Another is to use and place trackmarks and similar guides aright. The third is to recognize that the Trainz AI rules do not always reflect the real-world workings of signalling systems, dispatcher-routines and so forth. One must write the driver AI lists to suit Trainz, not proto-typical real railways.
On the other hand, the programming structures of Enhanced Interlocking Towers seems able to make Trainz consists and the signalling systems controlling them behave in real-world prototypical ways. Of course, the logic and language of EIT is rather more complicated that that of Trainz AI (or so it seems to me, having failed, so far, to master EIT programming conventions).
But others have managed it. For examples of both Trainz AI and EIT control of multiple consists intracting across complex junctions, try one or more of the various Dearnby & District routes & sessions.
Lataxe
The AI is nothing but mere 0/1 switches and is as good as the user made his/her route/session
so always go back to/check what YOU made, assume nothing.
EIT and any path rule/command system still uses the same AI, just some decisions are forced(pre defined)
Here i keep it very simple(KISS) only use very basic commands, but group them in libraries and use them with copy commands from
these building blocks are used over and over.
On my main route now 45 trains drive and its actually welcome that every 15min I have to "unstuck" a train
then think/puzzle why my train = I (not the AI) could not move on.
often changing a switch's default position or adding 1 signal, resolves it.
TRS19 (the original before any SP) is the best Trainz simulator i ever saw or used
its a very "open" structure and can be tweaked to meet all your needs.
Don't expect however that everything is done for you,
you need to spent time learning and discovering, falling/failing only enhances the joy for me
Once you can create content and scripts another dimension opens.
Find a version of Trainz, you like best, take time to learn
and just have fun (life is too short not to)
greetings GM
so always go back to/check what YOU made, assume nothing.
EIT and any path rule/command system still uses the same AI, just some decisions are forced(pre defined)
Here i keep it very simple(KISS) only use very basic commands, but group them in libraries and use them with copy commands from
these building blocks are used over and over.
On my main route now 45 trains drive and its actually welcome that every 15min I have to "unstuck" a train
then think/puzzle why my train = I (not the AI) could not move on.
often changing a switch's default position or adding 1 signal, resolves it.
TRS19 (the original before any SP) is the best Trainz simulator i ever saw or used
its a very "open" structure and can be tweaked to meet all your needs.
Don't expect however that everything is done for you,
you need to spent time learning and discovering, falling/failing only enhances the joy for me
Once you can create content and scripts another dimension opens.
Find a version of Trainz, you like best, take time to learn
and just have fun (life is too short not to)
greetings GM
I have several routes with AI drivers interacting with manual trains. One thing I have had to do is pay strict attention to the signaling - IT WILL NOT BE PROTOTYPICAL. But if setup correctly it will allow all trains to transverse the map without any problems. Some triggers and switch controls may be necessary - thanks to GM for helping me with that.
Here's just a few of the extras needed:
Trigger Multiple Signals - I use this many different ways. It is really good for controlling train crossings.
Junction Link - used to turn two junctions that are located close to each other
Set Cleared Junctions - move the points back to a preferred position because some AI don't reset the junction after passing through it
Trigger Check - Initiates more sub commands
Set Junctions - used with trigger check to turn a specific junction(s)
Directional Trigger Control - initiates sub commands based on the direction of the train
This is probably just the tip of the iceberg. There are nearly 600,000 assets in TRS19, so I'm sure I've missed a few. Interlocking towers was mentioned earlier. I have not attempted to learn the use of those, but one of these days I probably should.
TRS19 can do a lot and the learning curve might be a little steep, but it's a lot of fun and keeps me off the streets at night. Hang in there.
Here's just a few of the extras needed:
Trigger Multiple Signals - I use this many different ways. It is really good for controlling train crossings.
Junction Link - used to turn two junctions that are located close to each other
Set Cleared Junctions - move the points back to a preferred position because some AI don't reset the junction after passing through it
Trigger Check - Initiates more sub commands
Set Junctions - used with trigger check to turn a specific junction(s)
Directional Trigger Control - initiates sub commands based on the direction of the train
This is probably just the tip of the iceberg. There are nearly 600,000 assets in TRS19, so I'm sure I've missed a few. Interlocking towers was mentioned earlier. I have not attempted to learn the use of those, but one of these days I probably should.
TRS19 can do a lot and the learning curve might be a little steep, but it's a lot of fun and keeps me off the streets at night. Hang in there.
Since you appear to have a "solution" to the AI problems, perhaps you can explain why after setting a session, so it performs exactly as designed with smooth running, from end to end with several trains all performing complicated tasks, triggering other trains to perform other equally complicated tasks. And after you save and close the session, next time problems will start to occur. As per your suggestions we don't set up things the way they should. If so, how come the first time we run the session it works, but not after one or more saves? I have to say, IF N3V sees and READ this posts, when someone comes along saying that it is our fault for not setting AI properly, they will set aside our grievances and blame our inabilities, not the shortcomings of the program: Then nothing will get fixed. Fortunately there are more people complaining that praising AI. Regardless of how smooth you make it to work, it doesn't for me ( and many others like me), and most of us have quite a lot of experience to set and operate Trainz. At this point I dare N3V to come here and at least make a comment. At least, if you don't agree, I deserve it.
Tony_Hilliam
Trainz Plus - enjoy Trainz from just 20 cents a da
We don't read every post but we read "most". (And daring us to do something may work against you since it isn't very motivating for us.)
Saving is an interesting one since many older rules do not have any code that handles the saved state. So that would definitely be a game breaker in certain cases. The solution is to either update the scripts to support save or remove that script and replace with an alternative that does, or work around it completely.
Consider this:
The AI works by a set of rules. Your route is structured in a certain way. You give the AI a series of instructions and it goes through those rules and, for example, path finds to a destination, determines what speed to drive at, when to slow down for red or caution signals etc.
If ALL factors remain unchanged, the code will perform the same way every time, and the results will be the same. However, if ONE thing changes, then different results are possible. And that thing may be as small as a change of frame rate resulting in a slight variation in timing at a critical moment.
The AI isn't perfect, but it doesn't need to be, it only needs to be predictable. Your "job" is to work out how to instruct the AI to do what you want it to do. Invariably, another signal, a track direction marker or a waypoint along the way will resolve the vast majority of problems people encounter.
I would suggest people with problems post their routes and sessions to the DLS and invite other users to check them out and offer advice on how to resolve any issues.
Saving is an interesting one since many older rules do not have any code that handles the saved state. So that would definitely be a game breaker in certain cases. The solution is to either update the scripts to support save or remove that script and replace with an alternative that does, or work around it completely.
Consider this:
The AI works by a set of rules. Your route is structured in a certain way. You give the AI a series of instructions and it goes through those rules and, for example, path finds to a destination, determines what speed to drive at, when to slow down for red or caution signals etc.
If ALL factors remain unchanged, the code will perform the same way every time, and the results will be the same. However, if ONE thing changes, then different results are possible. And that thing may be as small as a change of frame rate resulting in a slight variation in timing at a critical moment.
The AI isn't perfect, but it doesn't need to be, it only needs to be predictable. Your "job" is to work out how to instruct the AI to do what you want it to do. Invariably, another signal, a track direction marker or a waypoint along the way will resolve the vast majority of problems people encounter.
I would suggest people with problems post their routes and sessions to the DLS and invite other users to check them out and offer advice on how to resolve any issues.
Tony..
With great respect. Do you really mean it when you say "daring us to do something may work against you since it isn't very motivating for us."??
Do you realize what you are saying in that statement? You appear to be saying that if a person "slams" you and is not "motivating" you have less of a reason to care for problems. As a professional, isn't the desire to have a great product, without flaws, all the motivation you need?
If enough people "complain" the N3V products are in some way going to be "less" than if people don't whine and complain?
Please make your products great because you want them to be great. "Motivation" should not be based on whether there are negative "vibes" in forums. Surely you agree with that?
Personally, I don't have a problem with TRS19. It has issues, but then any complex program (and this is nothing more than a computer program) has problems. I am a retired IT professional and Electronic Engineer. I've built very complex systems over the past 50 years and know that there is no way to make them "perfect". Fifty lines of code can be made to perform correctly 99.9999% of the time. One million lines of code with 250,000 "if statements", not so much so.
I have an issue with the power meter on some locos. It is incorrectly programmed to read backwards - i.e., full power is indicating braking. Geez! One of the solutions is make the power setting in the config script a negative number. That works great, but if the loco is payware, I cannot change it and I cannot make a clone of the asset for my own use. So, I'll just deal with it.
If you have a repeatable problem report it to the help desk. If it's randomly occurring, then there is likely other factors coming into play. Try troubleshooting the issue yourself. Don't expect everything to be handed to you on a silver platter.
I have an issue with the power meter on some locos. It is incorrectly programmed to read backwards - i.e., full power is indicating braking. Geez! One of the solutions is make the power setting in the config script a negative number. That works great, but if the loco is payware, I cannot change it and I cannot make a clone of the asset for my own use. So, I'll just deal with it.
If you have a repeatable problem report it to the help desk. If it's randomly occurring, then there is likely other factors coming into play. Try troubleshooting the issue yourself. Don't expect everything to be handed to you on a silver platter.
I think what Tony means is that he doesn't respond well to ultimatums
IMHO a lot (not all) of the criticisms of N3V in these forums are like armchair quarterbacks reviewing the Sunday Night Football game - "If I was Patrick Mahomes we would have won if I did it this way...". I find statements such as daring N3V to respond to a post (particularly an "armchair expert" post) or, as happens from time to time, "threats" to quit Trainz "forever" unless N3V fixes "this or that", all particularly childish.
I fully agree with Tony's statement about ignoring such posts.
I fully agree with Tony's statement about ignoring such posts.
If anyone was replying to my response 4 posts up...
Tony first line refers to "motivation," not ignoring posts and not giving to ultimatums. I agree 100% that Tony should ignore all the negativity. My point was this: Quite a bit of what's on the forums is negative. Too negative. But N3V's motivation, or lack of it, should not come from posts in the forums. That's all I meant.
And Yes, Tony should ignore the junk and demands. Absolutely.
There are some flaws and problems, yes. Yet, Trainz is terribly addicting, fun, and educational. There's your motivation! It's the product(s) that motivate!
Tony first line refers to "motivation," not ignoring posts and not giving to ultimatums. I agree 100% that Tony should ignore all the negativity. My point was this: Quite a bit of what's on the forums is negative. Too negative. But N3V's motivation, or lack of it, should not come from posts in the forums. That's all I meant.
And Yes, Tony should ignore the junk and demands. Absolutely.
There are some flaws and problems, yes. Yet, Trainz is terribly addicting, fun, and educational. There's your motivation! It's the product(s) that motivate!
Totally agree with you on all points. But how can we judge N3V motivation, or lack of motivation?
I spend my entire working career in an occupation where motivation was constantly being sapped from you by negativity almost all of which was from people who were not in that occupation/industry and clearly did not understand it. I sometimes wondered why I stuck with it but then I would remember that because I was on the "front line" (so to speak) I knew more about what was happening and what was needed than the many critics who never saw the whole picture and were mostly fixated only on what would benefit them. I often see the same thing in these forums - not blaming anyone as it is human nature.
The 2 most fateful phrases in the English language are, "I dare you" and "watch this" and usually spoken just before becoming a nominee for a Darwin Award. Wouldn't motivate me either
(The Darwin Awards celebrate those who spectacularly remove themselves from the gene pool.)
(The Darwin Awards celebrate those who spectacularly remove themselves from the gene pool.)
Last edited: