Trainz 2012: AI "clowns" are braking too hard at trackmarks

[JonMyrlennBailey]

This is occurring when I am trying to make my passenger train glide smoothly to a stop at the station platform. I tried putting in a 10 mph speed limit sign approaching the platform at either end. Changing the range setting of the track mark from default 20 m to 150 m doesn't help smooth out stops.

I set Rules in Session edit for Realistic control type only. I noticed as the train was braking hard at the track mark, the Cab control HUD temporarily turned into a DCC type HUD then back to a Cab HUD again after train had 'screeched' to a halt.

There is a rule called Vehicle Physics with an E=MC2 logo. Will this help?

I have also heard of embedding invisible speed limits. Are these available in lower speeds than 10 MPH. Can I use these to gradually step the AI trains down to an easy stop?

When these AI "drivers" hit the track mark it is as if they throw on the emergency brake, independent brake and full-throttle dynamic brakes all on at the same time full bore.

 

[cascaderailroad]

Drive manually in CAB mode, or DCC ... and you'll avoid all the AI, which are dumb as pigeons, following a detailed list of directions, and trackmarkers ... try placing many ever decreasing speedboards, in progression, before a station stop.

ie:

2 mile ahead: 60 mph
1 mile ahead: 40 mph
1/2 mile ahead: 20 mph
1/4 mile ahead: 10 mph
Right before a station stop, place a 5 mph speedboard

 

[JCitron]

The braking hard at the stations is the station script kicking in to guide the train to a stop at the "right place", open the doors, and do the right thing. They also do halt a bit more at the drive-to track mark commands. Have you tried the Drive-via or the equivalent navigate commands? They don't seem to stop then proceed, but instead slow down and then proceed on the other side once the mark is passed.

Being dumb as pigeons is an underestimate....

John

 

[JonMyrlennBailey]

The station I am using does not have its own track section built in. It is just placed along side normal track. I DO want my passenger trains to stop at this station where I want them to stop AND to gradually "creep" into the stopping point. It is a lot of guess work. I tried using the Speed-Signal triggers but it seems like those were instead for junctions. The lowest I can set these is 4 MPH. I have been monkeying around with them trying to figure out where to place them in relation to a track mark to get the smoothest-possible stop but it is like chasing a white elephant.

Apparently AI needs much improvement in following real-world physics.

 

[cascaderailroad]

Sometimes people expect so much perfection in a game, where trains run by automatic by robotic AI control ... and it just ain't gonna happen, unless you place many, many commands, and triggers, before a station stop.

Real railroad trains do not have AI automatic robotic control ... Why should Trainz ?

Real RR drivers, run a locomotive using manual control, moving the handles all by hand, following the conductor or dispatchers commands.

 

[JonMyrlennBailey]

You need to understand my purpose for AI operation. I plan to make a video of my train gaming to share with other people. I can use paid software called FRAPS to record my Trainz sessions. I will then burn these recordings onto Blu-Ray with a paid version of Ashampoo Studio 15 and I am hoping to impress others who watch my Trainz work. I can't be monitoring all my camera work and control all my viewing angles to make a Trainz movie while my hands are firmly glued to the loco controls. I will be too inundated with multitasking. I like to change camera modes from inside the cab, line-side, free-roaming, chase during sessions every so often. I want to be free to roam with my cameras while the trains do their own thing. I do want my trains to behave well for the camera however. It is just so hard to manually control trains and work the camera views at the same time. I have a lot of money invested in this whole gaming hobby. I have higher expectations than Mickey Mouse workmanship.

Real-world locomotives also have to obey real-world physics. They just don't stop suddenly like a rubber ball hitting a brick wall.

Trainz, after all, is nothing more than computer software. It should advance to the 21st century digital age.
Perhaps, those Trainz software engineers on the 5th floor need to go down to a rail yard and observe how real
trains operate or go to an Amtrak station and note how gracefully the California Zephyr eases to an exact point on the
platform. Obviously script controls trains speed and changes it at certain rate. I would think script would also tell
an AI driver exactly how far away the coming track mark is. So, let's say the train is moving 5 mph toward the
track mark 100 feet ahead of the train. What I want is a constant rate of deceleration from the 100-foot approach point
from 5 mph to 0 MPH right at he track mark. You do the physics math and geometry to determine the value of deceleration
in feet per second/second to make this so. This should be a reasonable rate of decel for a relatively-light passenger train
in the real world. A 10,000-ton freight may need more like 300 feet, a football field length, however, to decelerate smoothly to a stop from 5 MPH.

It seems like my AI drivers are going from 5 MPH to 0 MPH over the distance of only 5 feet, in real scale.

 

[Lots_Trains]

Ha

As self appointed President of the AI Drivers Union I am, and our many thousands of our members are greatly offended by the offensive posts above, please remove them at once, if not we are going on strike, in the meantime we will be working to the letter of the code, not one deviation. If you wish to have our membership to perform their duties properly, I suggest you fine tune you engine files and give us the proper equipment to work with, as President I have and I get perfect co operation for the members.

It should be noted that the members have a very low opinion of some of our employers.

Cheers

Lots

 

[cascaderailroad]

When making a film, you use Keyboard Commands:

Q: Release
A: Application
Z: Lap
W: Increase Throttle
S: Stop
X: Decrease Throttle
R: Reverse
F: Forward
B: Bell
H: Horn
L: Lights
V: Sand
Ctrl+D: Decouple

Run in DCC, free roam, and use trackside rotating cameras

AI will drive you batty, and every consist has a different enginespec, and difference in operation, so you will get completly different results every time.

It's not a perfect world ... and Trainz will never be perfectly realistic, especially with pigeon brained AI Drivers, who sit and wait forever at green signals, then stop, and back up a mile !

 

[Roy3b3]

Hey Jon, You mentioned that you wanted to focus on the 'movie-making' and rely on the trains to follow your instructions. Well, that's very easily organised.

To achieve some realism of a train pulling slowly into a station and coming to a halt at the end of a platform; you'll need to firstly make sure that the length of your platform matches the length of your train. Don't have more carriages than is needed. If you want a lot of carriages, then you need to extend your platform accordingly.
I've found by using 'AJS Invisible Stations' they can be modified to fit any circumstance and they slip right in under the station you have chosen. You can adjust your platforms from 50 meters to 500 meters long, set the height and shape of passengers to match the station or platform you are using. You can also use adjustable platforms to extend the length accordingly.

Set your speed levels just as Cascaderailroad suggested, to simulate how a real train would perform. Remember, Steam trains behave differently to the Diesels.

Here's a suggestion,... why not create a test-track of say two base-boards with two invisible stations (properly named) and an oval of track with some speed limits. Then position a few 'Markers' at 5 meter intervals towards the end of your stations to gauge exactly where your train will actually come to a complete halt. Program your Driver 'Set-up' to navigate to each named station with a 'Load' command following. Then give it a run, taking particular note at which marker the train runs to. This can also be handy because you can then add 'Hornz' slightly after the marker. This will activate as the train begins to leave the station heading for the next stop.

Lay out a few sidings that link in to your 'oval' with different trains, because you'll find each one could behave slightly differently to another. It's all trial and error until you get what you are looking for.

Good luck, cheers,
Roy3b3

 

[JonMyrlennBailey]

The trouble is, I need to be inside the cab or in chase view to monitor signal lights on my road. Rather hard to do in line-side view.
I also need to watch for grade crossings so I can blow my horn at the right time.

I have gotten some improvements with the invisible Speed-Signal triggers. I have to fiddle with their placement and speed values using the ruler. I found out the nose of the locomotive likes to stop at the outer set range of the track mark and not right on top of the mark itself. It is a juggling act for sure.

Here is my problem. If I put 4 mph speed triggers on my mainline right over the Tehachapi station platform, all AI trains are going to be creeping past this point at 4 mph and there are grade crossings nearby. A 4-mph 7,500 foot long freight will take 20 minutes or more to cross these grade crossings indeed and make my waiting AI motorists very impatient!

Can AI freight drivers be set to ignore the rules of the invisible stations? I only want passenger trains to creep in and out of stations slowly and freights to ignore station stop procedure altogether.

I just tried using an invisible station in Tehachapi. Won't work there as that station is on a 2% grade. There is no way to get the feet of the loading/unloading people flush on the ground with the sloping platform there. The platform is spline object and it can be sloped accordingly end to end. Invisible stations only work on level ground.

 

[nicky9499]

John, have a look into train priorities. I use them to differentiate which trains trigger certain sidings. The only "trouble" is you have the remember to set it appropriately prior to departure.

 

[oknotsen]

Look for "update max speed and priority" trigger on the DLS and play with the priority settings of the different consists.

 

[llebrez]

We are deviating from the original issue. In my experience, some locos stop abruptly while others do it smoothly. I has to do with the config of the engine. To verify what I am saying, run other locos into the station and see if everyone of them do the same. You'll see they don't. Probably what you did with trackmarks and speed limiters is right or an overkill.

 

[JCitron]

We are deviating from the original issue. In my experience, some locos stop abruptly while others do it smoothly. I has to do with the config of the engine. To verify what I am saying, run other locos into the station and see if everyone of them do the same. You'll see they don't. Probably what you did with trackmarks and speed limiters is right or an overkill.

I agree as that can contribute to AI actions, but I have found that there's a bit of random variation even with the same consists and AI, which makes some things impossible to always predict such as proper stopping and positioning. I wonder too if the close proximity of all the speed signs isn't confusing the AI. I've had an issues with this in the past with AI driving through a yard. I always had SPADs, not because of other driver flipping levers or going too fast, but instead due to the AI drivers stopping at each speed sign then surging up to speed again. This happened even if the next sign was the same speed. I eliminated the other signs, those that are in between the first and those in the station proper, and the AI moved along a lot smoother.

John

 

[JonMyrlennBailey]

Actually, here I how I got the best stopping action to a track mark, though not 100% perfect, the AI buffoons (V/Line A7 locos, Southern/ATSF F7 locos, N& W GP9P locos all pulling light passenger trains) all brake with the same level of hardness from 4 MPH to the track mark point.

I use a ruler from the outer range of the track mark (where the loco nose stops) as a starting point, X, then count the dots (hash marks) on the ruler including the arrows on either end. I place Speed-Signals at the following intervals to Point X as follows:

-9 dots from X, 4 MPH
-3 dots from above, 4 MPH
-same spacing, 5 MPH
-same spacing, 6 MPH
-same spacing, 7 MPH
-same spacing, 8 MPH
-same spacing, 9 MPH


Some (most) of these AI drivers are idiots as some like to actually SPEED up after passing the LAST 4 MPH invisible speed-signal trigger just before the track mark if it is too close or too far away from the track mark. I noticed this after an hour or so of trial and error speed trigger placement so sped trigger placement is critical here.

 

[Bill69]

You have to be very careful using multiple speed signs to slow a train. If the last vehicle in the consist passes the 9mph sign after the loco has passed the 4mph sign the driver will think the speed limit is 9mph.

Also in yards the default direction of the junctions is crucial. All junctions should default to the yard main branch when nothing is happening there, then when a train has to enter the yard it only has to switch one of the junctions for the track it has been ordered to instead of all the junctions before that track. Likewise when the train is leaving the yard it only has to switch the junction for the track it is on and possibly the main line junction.

Once you understand the AI properly it is amazing what can be done, though having said that it is more difficult in TS12 than it was in TRS2009.

Cheers,
Bill69