Page 5 of 5
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Sep 14, 2010 1:36 pm
by MidnightTrain
pilot37 wrote:I found the Trainz 2010 AI quite good considering that I made my own route and kept adding perpetual trains to it. But in the end the program could only take so much...it became a game of "see how many trains I can keep running before it all comes to a halt" A long time ago I used to program programmable logic controllers for industrial machines. You had very few instructions you could use so the logic solutions got to be quite a lot of fun. It was amazing how the programs needed tweaking every now and then as a machine found yet another new circumstance and crashed to a halt. But it would certainly suggest to me that with more complex language a better program could be developed for Railworks train routing than we have.
Railworks is all about immersion and atmosphere for me. Let's face it, there is not much to train driving, even with the steam locos. To do the job properly and protect itself against newcomers I would be hoping for the following in order of priority:
1. Stutter fixed
2. Scaled up Coal or Diesel consumption so I have to manage fuel.
3. Better weather effects...raindrops on windows and frost on glass, heavy snow, howling wind etc...
4. Real time lighting effects for time of day (Scaleable), working headlights.
5. Better AI which adapts to changes of plan (like being behind schedule, train derailments etc.)
6. Perpetual trains (eg loading and delivering coal on an ongoing basis)
7. Programmable Industries, supply and demand (Now I am asking too much?)
I would be happy to pay for some of this....for example in Flight Sim X I have "Real Weather Extreme"
I agree with the following, 1. 3. 4. 5. 6. I don't agree with 2. because I think the focus should be on
realistic fuel consumption. I drive mostly modern diesel locomotives, and they are highly fuel efficient machines. I don't agree with #7. because I think that goes beyond what is needed for this simulation.
As for AI in Modern US freight train operation, I do think RSC.com need to read up on ABS and CTC.
http://en.wikipedia.org/wiki/Automatic_Block_Signal
http://en.wikipedia.org/wiki/Centralize ... ic_control
There's also TracK Warrant and Direct Traffic Control
http://en.wikipedia.org/wiki/Track_Warrant_Control
http://en.wikipedia.org/wiki/Direct_Traffic_Control
Hopefully the following information will make a better AI and Dispatch program for US freight railroad that is in RW.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Sep 14, 2010 2:42 pm
by pilot37
I think where I am coming from on scaled up time and fuel consumption is that I don't have enough spare time or inclination to drive for very long periods of time so I would like to condense the effects. For example (as I mentioned in another thread) If I want to simulate a 24 Hour Le Mans race I can do it on rFactor, I speed up time, tyre wear and fuel consumption and do the whole thing in 4 hours. Without the flexibility the sim gives me I would never experience the sun setting, rising, and would only make 1 pitstop.
Is the answer to selling to diverse groups of customers with highly individual taste, to build in choice and flexibility. If Railworks doesn't do it, somebody else will and the sim will have a shorter life....bad for everyone (especially with the expense
)
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Sep 14, 2010 3:10 pm
by MidnightTrain
pilot37 wrote:I think where I am coming from on scaled up time and fuel consumption is that I don't have enough spare time or inclination to drive for very long periods of time so I would like to condense the effects. For example (as I mentioned in another thread) If I want to simulate a 24 Hour Le Mans race I can do it on rFactor, I speed up time, tyre wear and fuel consumption and do the whole thing in 4 hours. Without the flexibility the sim gives me I would never experience the sun setting, rising, and would only make 1 pitstop.
Is the answer to selling to diverse groups of customers with highly individual taste, to build in choice and flexibility. If Railworks doesn't do it, somebody else will and the sim will have a shorter life....bad for everyone (especially with the expense
)
Yea, it would be good for RW to give us the option to either speed up or slow down time. Personally I like playing the game in real time, it's much more realistic. However, different strokes for different folks.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Fri Sep 17, 2010 1:16 pm
by USRailFan
I suggest that everyone play the "Heavy freight" scenario on the Hagen - Siegen route (and then I mean really play it, to the end) . Then you can tell what you think of RW AI and signalling logic. Especially compared to the youtube vids shown in the initial post.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Fri Sep 17, 2010 3:17 pm
by pilot37
USRailFan wrote:I suggest that everyone play the "Heavy freight" scenario on the Hagen - Siegen route (and then I mean really play it, to the end) . Then you can tell what you think of RW AI and signalling logic. Especially compared to the youtube vids shown in the initial post.
Guess you must be a school teacher or something?
Enlighten me, is it good / bad or just compulsory?
I have put a lot of time in on Railworks but like to do my own scenarios, I am getting better at it all the time. I am neutral on AI, there are more important things to me as per my previous priority post (personal taste / opinions only).
If Railsim.com are passionate about their product then they will naturally improve it as they have done in stages so far (scenario tools really helped big time). If they hit a brick wall then they will know it. I can only make one strong suggestion to them...communicate better with customers. They are very fortunate to work in an environment with such blunt feedback. In my line of work it is difficult to get honest feedback.
I personally feel very optimistic, I love the product and believe it will get better, both core product and addon features. I worry that the stutter is a brick wall, hence no feedback.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Fri Sep 17, 2010 3:24 pm
by USRailFan
OK, short description:
You start the scenario in a light engine, then go over to one of the nearby sidings and pick up a rake of tank cars. After a while, you get permission to leave the yard, following a regional train. The scenario then runs OK for 10 minutes or so, when the regional train you're following gets a red, and then sits there (and your freight train sits there too obviously, on the next signal block) for _one hour_ watching first a freight train, then a light engine, then another freight train, then a regional train, and then a third freight train go by, before the regional finally gets a green. Then bear in mind that this is in tightly populated central Europe, the scenario is set in the beginning of the rush hour, where passenger trains have absolute priority over everything else, run at hourly or even half-hourly intervals, and certainly would _not_ be made to wait for _one hour_ outside one of the route's bigger stations... Maybe five minutes, perhaps 10, MAXIMUM. Also, from the (very brief) info given at the start of the scenario, it is pretty obvious that you're intended to 'whiz through' most of it... How this scenario even got past the alpha testing stage is beyond me...
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Mon Nov 29, 2010 6:17 pm
by trainsmit
It's now a while back in RW1 - and for various reasons some time since I've driven behind a "crawling" train, but back when, I one decided on a little experiment:
Sneaking up behind the crawler - SPADing a signal - I then lowered speed to 2-3 MPH until contact and then simply pushed the crawler past the next home signal. As the track ahead had meanwhile been cleared, the former crawler picked up line speed and vanished in the haze. This might still be workable. This also work with freighters, that the heading loco cannot pull over the ridge - allowing you to drive a pushing service.
trainsmit
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Mon Nov 29, 2010 7:05 pm
by ianb1234674
trainsmit wrote:It's now a while back in RW1 - and for various reasons some time since I've driven behind a "crawling" train, but back when, I one decided on a little experiment:
Sneaking up behind the crawler - SPADing a signal - I then lowered speed to 2-3 MPH until contact and then simply pushed the crawler past the next home signal. As the track ahead had meanwhile been cleared, the former crawler picked up line speed and vanished in the haze. This might still be workable. This also work with freighters, that the heading loco cannot pull over the ridge - allowing you to drive a pushing service.
trainsmit
I've done this on the Bristol-Exeter route went a bit too fast though on the venture scenario. I wish they could do this for real when the trains are running late but I think the passengers might complain a bit.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Mon Nov 29, 2010 9:35 pm
by malkymackay
Now and again you see a picture that shows a perfect example of what someone wishes could happen
http://web.ukonline.co.uk/53amodels/pic ... 250478.jpg
As an aside, for anyone who says superelevation isn't noticable, look at the above image again
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Mon Nov 29, 2010 9:47 pm
by bristolian
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Nov 30, 2010 12:08 pm
by Kromaatikse
I'm pretty sure I saw mention of a Jubilee on a railtour being ordered to assist a failed 50 on a freight working, back in BR blue days.
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Nov 30, 2010 12:42 pm
by bigvern
The main problem these days is that not many traction types will couple to anything else, unless you have an adaptor coupling that looks like something out of a medieval torture chamber. Then there's so many electrical controls running through the BSI coupling on sprinters etc., chances are if you do attach another unit the fault transfers and you end up with two stuck trains.
It was much easier when everything had compatible couplers!
Re: Am I alone in thinking the AI needs much more "I"?
Posted: Tue Nov 30, 2010 3:10 pm
by exshunter
Nice pics Bob, i went to school @ St Thomas More and had a pretty good view from there! The memories come flooding back
