Are 9 AI drivers too much?

martinvk

since 10 Aug 2002
In build 100240
Built a simple test map, a central loop and four arms in a star formation around the perimeter. Started with 4 AI consists, one for each arm that goes around the central loop, stopping at one of the four stations and then out and back on its arm. Added an inner loop with 4 AI consists going around. No problem so far. All the AI's interact and politely wait if another consist is in front of them until the way is clear. Now I added a 9th consist that visits each arm in the star formation in turn as well as the central loop stations.

Now one of the AI consists will suddenly not continue. The track is clear, the next command is clear but it refuses to move. If I switch to manual DCC and turn the speed dial, the consist will advance but as soon as I return to the AI schedule, it stops.

Another anomaly is that each consists on the outbound leg will suddenly slow to a crawl, as if it was entering a station and preparing to stop to load passengers. Then it will resume and return to the normal track speed. All consists do this.

If I abandon the schedule for the stuck consist and issue a Drive command, all is well, even the slow to a crawl issue disappears. It seems the schedule command interpreter is causing giving my AI drivers a hard time. Are 9 AI drivers too much? I don't remember having this problem before.

It's a simple map, only some track, four stations, signals to control the trains, and a few track marks and direction markers.
 
Hmmm, removed one consist, now there are only 8 doing their thing, following a schedule and so for no glitches. Still have the annoying slowdown. Funny, the arms are all different lengths but the slowdown always occurs near the end, just before the return. Will have to try a different track configuration to see if that makes a difference.
 
I have around 12 drivers and they generally behave. But I have had the same experience as you with slowing down before a station they were not scheduled to stop at. No fix that I have found. As far as drivers stuck, beyond the obvious broken track and misplaced signals, I have found more subtle issues that cause problems:
1. Junctions missing their switch lever
2. Switch levers where there no longer a junction
3. Pasted stations. What I mean by this, let's say you have a town with a station that you want to relocate, so you copy and paste the town to its new location. Even if you delete the original town, the AI driver will randomly pass through the pasted town's station without stopping, but not recognize they went through it and then continue to search for it until its totally confused and stuck. Solution: Rename pasted stations to something else and make the session changes.
 
Martin,

I've had that with yard setups where the loco will creep up to a consist if there is a signal facing towards the parked consist. With that said, what is your signaling like? That maybe what's causing the issue you have.

schweitzerdude,

That has always been a problem. If an industry or track mark is relocated after adding it to the schedule, either will be ignored. Deleting these from the schedule and adding them back in seems to fix the problem.
 
A quick update. On a whim, I deleted the existing guides where the slowdowns were occurring and replaced them with the same guides. Low and behold, no more slowdowns. Wish I knew what I did wrong the first time.

For those curious, this is what the map looks like so far.
TRS 2019 100240 map.JPG


This is the guide that I used to align my track, 2t 4m kuid 70337:23083:2 The one on the right is for illustration only. What it looks like before any track is attached.
TRS 2019 100240 guide 2t 4m.JPG


Here is an overview of a typical station. In the foreground, a right diverging signal LHS kuid 49267:24015 in the center and an absolute signal kuid 49267:24006:1 on the right. The yellow buttons are 4m spacers kuid 70337:23015:2 to help space the track visually. The dark halos help align catenary wires if any are used. In the background two 162m platforms with passengers kuid 70337:23173:12 On the ends of the platforms, low platform signals kuid 49267:24035:1
TRS 2019 100240 signals and platforms.JPG
 
Back
Top