Odd behaviour of AI.

[seagoon]

I've been running a couple of Activities under OR and have found that - as far as I can tell - those that have the AI running on parallel lines, where there is no requirement for them (or the player) to cross each other, seem to run okay. The player passes the AI heading in the opposite direction and all is fine.

But when there is single track involved, and stop signals and one having to stop for the other, some odd things occur.

The most recent that I've been experimenting with is one from MLT's "Soldier Summit".....'Castle Loadout 2' is the Activity name.
When unmodified - that is, exactly as it would be run under MSTS - I have a situation where one of the AI approaches a stop signal, slowing down as would be expected. But once it stops, it then reverses and reaches a speed in excess of 200 mph before disappearing into the great unknown somewhere back from whence it came.

But when I modified the "Expected Performance" from the preset 75% down to 50%, the behaviour improved: the AI train kept to the maximum speed limit, reached the stop signal and stopped. Well....it actually stops a bit further past the signal, but at least it didn't do a warp-speed reverse.

Dropping the Expected Performance down to 30% had the effect of the Activity "timing out". A pop-up window appeared with some info in it but whilst I was trying to read it the Activity terminated and I was dropped back to the OR menu.

Has anyone else experienced similar behaviour?

BK

[csantucci]

Interesting observations. I saw AIs running away at ever increasing speed in my activities at their reverse points instead of reversing, but only when they are "out of sight". When they are seen on the graphics, I found out they behave better. However longiron told us that AI working at the moment is very provisional. I expect that things should improve significantly for next release, where signalling should work properly.

Carlo

[seagoon]

Hi, Carlo.
Yep....what you have described is exactly what I've experienced since my initial post....and it also happens even when I adjust the expected performance.
It's all part of the development, I know and understand. I just thought I'd report it in case it helps.

Another oddity which I've experienced is that with some of SLI's locos, once the dynamic brake is increased past "Setup" it is impossible to reduce or increase the effect and impossible to get out of the dynamic and back to throttle.
This happens with their D9, ES44 and SD70MAC so far.
I haven't experienced it with any of MLT's equivalent locos.

BK

[longiron]

The ORTS team is aware of the AI issues. We are in the process of rewriting the entire signaling and AI software logic. It will be available in the next public download.

[seagoon]

The ORTS team is aware of the AI issues. We are in the process of rewriting the entire signaling and AI software logic. It will be available in the next public download.

No worries, Chris.

With regard to the dynamic brake problem I mentioned.....has that been reproted by anyone else and should it be mentioned to SLI?

Bruce

[longiron]

No worries, Chris.

With regard to the dynamic brake problem I mentioned.....has that been reproted by anyone else and should it be mentioned to SLI?

Bruce

Bruce,
We're are aware of the issue. It's a backward compatibility effort for the SLI throttle configs as well as some others that are used in European routes. It's not likely we will get to this enhancement in the next release, because there is a temporary workaround.

[seagoon]

Thank you.

BK