AndiS wrote:Keith, I never tried loading instructions with AI, but your example with the screen sounds like each of the wagons should get "load" instruction for the bulk loading facility there. Then the AI should stop for some time for each of the wagons in turn.
Like Pete said, automotive wagons are just any wagon plus any engine blueprint where you set maximum speed to a few mph and maximum force to something smallish to prevent rocket starts.
Would it be any easier if we had many single, self-propelled wagons? What I mean is that, with individual wagons you could have a very short marker under the loading point over each track so that the wagon was forced to stop at the correct position. Each wagon would have its own set of instructions so there wouldn't be any need to invent a way to have a sequence of loading for a short train.
If the iterative loading of an AI train does not work as planned, you would need to resort to the scripting scheme that Pete suggests. But aligning with each of the wagons would need extra care and altogether, it could become quite some project. And remember that AI does not receive the signals Pete mentioned. So the AI train would need a series of stop-at instructions, spaced by a wagon's length. But then, the "signal" could start the loading animation whenever it sees that a train stopped under the screen, without any sending of a signal. But the rise of the load level could not be implemented. This is a property of the wagon and if it does not receive messages, it will not cooperate.
That seems too complicated to be readily understood by non-technical scenario writers. I was hoping for a solution that was fairly simple to implement.
It's a pity that launching from a portal is unreliable. If it was easier to use you could have two portals underneath the screens on each line, with the first eliminating the unloaded wagon and the second launching a loaded one a short time afterwards.
[/quote]Regarding the co-existence of AI and player trains, you need to try each case. I have seen many cases where AI panicked and fell into 0.5 mph zombie crawl although there was really no reason, because there were sufficient signals between player and AI.
So if there is a rake of tracks where the player shunts, a nearby AI consist inching forward under the screen may or may not panic. It seems that they panic less if you approach them from the rear. What they hate is a player in the 6 or so blocks ahead of the AI engine, even if it has orders to reverse and never go where the player is. It may help to set the (final) destination of the consist under the screen to the end of that track. Then even the AI dispatcher should see that no one crosses the path of the AI train.[/quote]
Scenarios with a lot of AI can give very peculiar results. I tried to develop a long (~3 hours) scenario for Settle & Carlisle v7 where the player drove a 4F on a pick-up goods from Hellifield to Carlisle. I'd set up the AI in the scenario in the direction of travel so that I could check that the trains passed at the right spot. The first symptom that all was not well was about an hour in when after which approaching AI trains showed red lights instead of white. By the time I got to Appleby things were completely out of control. Trains would stay at their starting position long after their departure time, then vanish and appear where they should have been at speed. Thins got increasingly odd to the point where I binned the whole thing long before I got near Carlisle!
Keith