Open Letter to OR Team - AI/MP/Signals

Collapse
This topic is closed.
X
X
 
  • Time
  • Show
Clear All
new posts
  • SurvivorSean
    3000 Post Veteran
    • Apr 2001
    • 3460
    • Welland, Ontario, Canada.
    • General

    #1

    Open Letter to OR Team - AI/MP/Signals

    Hey guys I would like to personally applaud your efforts as I really think this simulator has the potential to be something that train simulation should of been years ago. Many simulators have appeared on the scene since MSTS started and even before it did. But through this whole process I think all these simulators missed out on what the true value of what railroading is about and that is more than just rails and machinery. In fact either in it's early days where it took 4 man crews and massive array of signal towers, to todays 2 man crews, automated units, huge CTC territories, etc. regardless of era it takes people.

    What I mean by that is despite most train simulator projects attempting to create realistic signal systems, and AI behavior. Not to mention random factors and purpose (which goes hand in hand with operations) they all require people.

    I gave up on the route editor a long time ago, and despite trying out Trainz the end did not justify the means. Many routes are built as prototypically correct and in some cases intermediate signals are used just like in the prototype. In fact a number of routes have been popping up over the years that are so realistic that in some territories like ABS and track warrant absolute signals are gone all together.

    The signal system in MSTS is seriously flawed for error. It does exactly what it's supposed to do according to the scripts. Checks the signal clear ahead value and tries to automatically dispatch. The problem is the clear ahead value doesn't work for intermediates as the logic behind this is intermediates or approach signals only appear on safe following movements and not at control points. To further complicate the problem AI services approaching a restricting signal will never regain it track speed as this is a bug that was never resolved. Another bug is competing for tracks as if a train comes up to a situation on startup with enough reverse slack, or if you dare reverse your train without a proper RP you may wind up giving up your player path for an AI service already spawned miles away.

    Now there are tricks to getting around this and some route developers are aware of this. Some routes work better than others. Cookie cutter type routes work the best as the nodes remain even in passing paths and appears to handle things well. But the dispatching system on multiple track lines never take advantage of the multiple tracks and crossovers that they are intended for. For example a faster train approaches a slower train with no opposing traffic would often get routed around this. But everything is so set-up in stone that there is little sense of excitement or change. Random factors such as those shown in TD3 barely work even with software like Activity Changer because MSTS is not able to adapt to changing situations.

    For the time being except for diversion I've abandoned MSTS. I'm in the process of creating a multi-player TD3 environment that will work. After that I'm going to look into having the dispatched territories generate proper paths with waiting points automatically through my RTS software that is in development as well. As such OR is a ways from getting there as my RTS is in the same boat.

    What I'm hoping is that while I develop the true team aspects of dispatching and AI through TD3 and eventually MSTS we may meet along the way. My programming expertise is way behind by at least a decade and I have a strong learning curve ahead for RTS which I refuse to write in any form of C. So I wish you all the best of luck and hope you'll take my thoughts into consideration and I hope to meet you on the other side.

    Thanks

    Sean
    Discord https://discord.gg/zfxTBbnu3K
    YouTube https://www.youtube.com/c/SeanMurrellRTS
    sigpic
  • sniper297
    9500 Post Veteran
    • Nov 1999
    • 9545
    • Union Grove, Wisconsin, USA.
    • General

    #2
    And fix the cigarette lighter!!

    Primary purpose of the PO&N with its open ended or "freestyle" activities was exactly what you're talking about, a compromise to recreate realistic operations. In real life trains don't adhere strictly to a rigid precast schedule, if they could then they wouldn't need signals or dispatchers. Most people seem to be content with scenarios/activities that are rigidly timed and cast in concrete before the activity starts. If the 3:10 from Yuma is scheduled to go thru Crowe junction at 5:15, Extra 2099 is supposed to wait until it passes, everything is fine provided the 3:10 is on time. What if it's held up somewhere and is 3 hours late? is every single train on the schedule supposed to delay 3 hours to match? Not in real life, the dispatcher would clear Extra 2099 to go earlier and shuffle the others around for the least overall delay. That's the part of railroading that's most interesting to me, the idea is to play the same activity over and over with VARIETY, what the AI traffic does and where you meet it depends on how long it takes you to do your switching tasks.





    That system is fragile but it DOES work, if you park your train on a single track section for 30 minutes you'll back up AI traffic to both ends of the line - but once you clear the section the AI traffic sorts itself out and gets moving again.

    So what you're looking for is already doable in MSTS, but it does need some tweaking. One feature that might be worth looking into is an ingame map that takes you into "dispatcher mode", where you can issue simple orders (stop at YB and proceed after my train passes, disregard previous order to take the hole at AZ, instead proceed to RF and wait for AI51 there) to untangle traffic jams, then hop back in your train to continue. A lot more fun AND more realistic than having everything locked up before you start.
    sigpic

    Comment

    • SurvivorSean
      3000 Post Veteran
      • Apr 2001
      • 3460
      • Welland, Ontario, Canada.
      • General

      #3
      PO&N is a great work around for MSTS. I've had plenty of fun with it in the past.

      Many people are either fine with running their trains in explore mode, freestyle activities, or managing with activities someone else wrote on prototype lines. For me it gets old very fast and from the outside I may appear like a loud minority, but I believe many others shelved MSTS a long time ago (and didn't find much better luck with any other simulator either).

      PO&N is believable from a short line perspective as there wouldn't be any other AI traffic on the parts the player switches. There are other work arounds that work fine with AI traffic interacting with player services if the route is set-up properly.

      But workarounds aside the majority of mainline operations involve you running a train that interacts with other services on the same lines. I like the pop up dispatch idea as that would be a tremendous improvement. I was once going to write a route like PO&N based on an area like Fostoria, Ohio where mainline trains kept running back and forth and a short line interacted with it on their own tracks. But I've never been fond of the route editor and was too much of a perfectionist so I avoided it all together.

      I think the key to realistic dispatching is going to be to involve real people which leads to multi-player. Even the automatic ATC functions I would imagine have their limits in Flight Simulator. But I do know VATSIM works like a well oiled machine for multi-player on their platform. Will there ever be a VATSIM like system on every line world wide for train simulators, highly doubt it. But I do see certain railroads and regions having the opportunity to become 24/7 operations. But right now only the dispatching will be real time because there is no simulator on the planet that can truly run a train simulator in real time multi-player unless of course you run in explore mode and react to invisible trains.

      So I'm going to let those that know C programming do their thing. Those that know the physics of engines and how pretty the trains should look keep up the good work. But in the mean time I've found my purpose and it starts from the top down. A live multi-player dispatching system is in the works and should be live by the end of the summer.

      Once I've worked out the kinks in the manual system I'll introduce RTS in managing traffic and other factors and some how build up a framework for the dispatching side of operations to be available for the use of other simulators. If those other simulators aren't capable then my next step is the nutty task of writing a path and service generator for MSTS along with waiting points, DRP's, etc. Even that with MSTS is no guarantees but I hope that OR is going to succeed in doing something different than all the rest.

      Thanks

      Sean
      Discord https://discord.gg/zfxTBbnu3K
      YouTube https://www.youtube.com/c/SeanMurrellRTS
      sigpic

      Comment

      • conrail1
        Senior Member
        • Jun 2010
        • 786
        • NY

        #4
        Originally posted by SurvivorSean
        I think the key to realistic dispatching is going to be to involve real people which leads to multi-player.


        The key to “realistic dispatching” is the same key to realistic railroad operations (including train handling) in general, getting individuals with actual experience in the railroad industry involved in the development of OpenRails.
        Asus Maximus VI Extreme Intel Z87, Intel Core i7-4770K, Corsair Dominator Platinum 16GB PC3-19200 (2400MHz), EVGA GeForce GTX TITAN SuperClocked, Samsung 840 Pro Series 128GB SSD (Windows 8.1 Pro 64)/Samsung 840 Pro Series 256GB SSD (Programs), CaseLabs Merlin SM8 with Corsair AX1200 PSU

        Comment

        • SurvivorSean
          3000 Post Veteran
          • Apr 2001
          • 3460
          • Welland, Ontario, Canada.
          • General

          #5
          Well the good news is you guys certainly have that because Open Rails is built by railroad enthusiasts.

          The problem that you will come up on that others have faced is trying to automate the dispatcher. We all know about ABS signals and of course MSTS and others already can do this just fine with simple scripts.

          If I recall I've heard of automated dispatching in Europe as well as I believe in the south west United States. I remember someone saying they were at track side and watched the switches keep changing from reverse to normal several times as 2 trains approached each other for a meet.

          When I personally dispatch I try a logical approach that goes something like this. I use a 30 minute cycle that overlaps itself every 15 minutes. I know I mentioned this recently but I don't think it was here so I'll repeat what is involved. This is an example I use on the busy CSX SB for TD3 where at times as many as 30 trains can be on the territory at once.

          Every 15 minutes I go into the arriving trains and look for anything that is going to arrive in the next 30 minutes. Provided that there is no conflicts with any traffic in the immediate area I'll either clear the 1st signal or make a note.

          I then start at the northern or southern end of the territory and work my way through the trains to the other side. On the hour and half hour I start at the top (north end) and on the 15 and 45 minutes I start on the bottom (south end). All trains on the territory should already have some clear track ahead of them, but for each train I attempt to clear 30 minutes. If I have to I stack the signals (this is a method used by dispatchers to have a line automatically clear once the previous train gets out of the block. A stack as the name implies can become quite complicated and involve several clearances at either the most basic or complex of junctions if traffic warrants.

          When traffic is moderate this works quite well. By the time he is through the cycle he should have just under 30 minutes (less the time it took to set up the routes, go get a coffee, pee, and get off the phone or radio with other employees). What's key about 30 minutes is it overlaps the next cycle that begins every 15 minutes. The more your pressed for time the less attentive you have to be as you approach the other side of your territory because you will start from the opposite side.

          The good thing about TD3 is despite some of it's weaknesses it is very good at random factors. Such random factors can include more simple factors like the train's performance, startup delays, and arrival delays. But it goes even further when you throw in switch failures, signal failures, hot box detectors (both false alarm, and actual alarms), engine failures, broken hoses, etc. The simple delays alone make precise timing go out the window and require a wait and see what happens. But the more serious delays can throw everything into chaos.

          If everything is as cookie cutter as these simulators make things out to be there would never be an issue. That is why MSTS some times runs better on certain routes than others. For me cookie cutter routes and activities is not what we really want. We want trains to arrive late once in a while, we want a real hot box detector, a signal failure, a coupler breaking (not the kind that always happens in MSTS when something is wrong and boots you out). If were sitting at a red signal we want to not just ask permission to pass it, but actually have the dispatcher give us a signal we can use. We want dispatchers that can also call the crew up and take away a signal if they are delayed or slower than expected.

          Hank and others have come up with some great signal scrips that I feel do a great job with following prototypical rules. That is not the issue with MSTS. The issue is it doesn't have the AI capable of adapting to change. The closest thing that comes to this is passing paths and I believe this was a feature that wasn't fully developed to what was intended.

          The point I'm trying to make is this. The prototype would love to automate the process but often still requires manual intervention. As much as I try and follow a routine with dispatching CSX SB I often find I have to make an exception the rules and at some point things jam up with so many variable circumstances that even a system with allowing you to become a dispatcher for a minute may get you into more trouble especially if the AI dispatcher had something else in mind.

          So in a couple of months a group of dispatchers will be simulating a living railroad. I know this has been done in certain groups before (nothing new) but I intend on taking it into the V scale realm. Even if all I can get out of it is a path/activity generator from TD3 logs at least I'll get something out of it.

          The GTA/NC/Bala from Maple Leaf Tracks has become my baby. It's close to my home and despite a little bit of protolancing it should serve about half a dozen dispatchers at one time all working together to run a railroad. Then with that purpose in mind I will reach out and see who wants a piece of that pie. I say let those that just want to run a train, run their trains. But I think those that want to operate a railroad may be interested in what is coming down the pipe.

          Thanks

          Sean
          Discord https://discord.gg/zfxTBbnu3K
          YouTube https://www.youtube.com/c/SeanMurrellRTS
          sigpic

          Comment

          • conrail1
            Senior Member
            • Jun 2010
            • 786
            • NY

            #6
            Originally posted by SurvivorSean
            Well the good news is you guys certainly have that because Open Rails is built by railroad enthusiasts.

            As we’ve seen in the past with every other train game, “built by railroad enthusiasts” just doesn’t cut it.
            Asus Maximus VI Extreme Intel Z87, Intel Core i7-4770K, Corsair Dominator Platinum 16GB PC3-19200 (2400MHz), EVGA GeForce GTX TITAN SuperClocked, Samsung 840 Pro Series 128GB SSD (Windows 8.1 Pro 64)/Samsung 840 Pro Series 256GB SSD (Programs), CaseLabs Merlin SM8 with Corsair AX1200 PSU

            Comment

            • SurvivorSean
              3000 Post Veteran
              • Apr 2001
              • 3460
              • Welland, Ontario, Canada.
              • General

              #7
              Big difference this time around and you have to take a look at the big picture to understand why.

              I am a railroad enthusiast. Now being that doesn't mean that what I think a railroad simulation should be will magically appear like the Easter Bunny but I do know what I want to see.

              Now take a good look at what makes a project happen. 10 years ago there were enthusiasts doing the best they could with limited resources creating such things as text simulators, 8 bit graphics that was slightly nicer than a Commodore 64. Still at the time it was pretty good, and fun to try.

              M$ for years supported a Flight Simulator franchise and they ventured into the world of Train Simulation with MSTS. Now despite it being a success, because it's a small blip on the Simulator market, and barely visible spec on the entire game industry it could only go so far. The funny thing is if Microsoft did further research they would probably understand that if they got it right and spent the amount of time to really do a good job and bring out MSTS2 they probably could of sold it for 99.99 a copy. Here's what M$ doesn't get, if people are willing to fork out dough for model railroading, if they managed to capture what model railroading was all about and combine it with what MSTS brought I seriously think they could of pushed V Scale to the limits that may have shown an impact on some of the model manufacturers balance sheets.

              Even if Train Master could understand what was out there just waiting they could of continued to focus on the high end users for their pro version and still see a viable income in their consumer version. But just because you have a bunch of rail enthusiasts who are programmers does not equate for a project to see the light of day for one reason. These enthusiasts don't control the money in the business. Companies and shareholders do.

              Now take a look at the Open Rails project. You might say so what, it's incomplete, has bugs. What did we ever get from M$ or any of these other simulators that became vapor ware? Hey we got nice videos and screen shots. Don't forget all the bios about how real rail enthusiasts were on board, but it was corporations that pulled the plug. At least we have something to sink our teeth into and a road map that seems to be plausible.

              If I had the money I would hire the programmers myself and start a company. Some could say why pay this when we have this etc. but I believe in the laws of supply and demand it would be worth every penny. Right now everything I'm doing is a hobby and because it doesn't pay the bills I'm doing it at my own pace. But if I ever get the momentum going nobody will pull the plug because I will be calling the shots. If someone beats me to it so be it. In the meantime I'll have a run for my money simply because like every other product out there an open concept project is in the works and enough people ware willing to get it done for free, and I applaud them for that.

              If a product does come out and makes more money per unit making it profitable you can bet the big boys will take notice. The competition will only make the industry grow and get better for it. If I pay money for dues to a model railroad club, you can bet I'd pay a small fee to be part of a multi-player network. But the thing here is despite any ambitions I may have to make some extra cash at enjoying what I do there is one thing that trumps all of it. Open Rails is being done as a volunteer effort with no strings attached. They're already making tremendous progress and it's only going to get better with the passage of time. I'm not an Open Rails team member as I'm taking a different path but I wish them the best of luck. All of these advancements will only enhance this hobby and the people with the bucks that could care less about our hobby can't do a thing about it except that they missed out on an opportunity that is simply going to pass them by.

              Thanks

              Sean
              Discord https://discord.gg/zfxTBbnu3K
              YouTube https://www.youtube.com/c/SeanMurrellRTS
              sigpic

              Comment

              • jovet
                3000 Post Veteran
                • May 2006
                • 3050

                #8
                Originally posted by SurvivorSean
                The funny thing is if Microsoft did further research they would probably understand that if they got it right and spent the amount of time to really do a good job and bring out MSTS2 they probably could of sold it for 99.99 a copy.
                I would pay hundreds of dollars for an exceptional train simulation software.

                Comment

                • cp5513
                  Senior Member
                  • Jun 2006
                  • 419
                  • K-W, Ontario, Canada

                  #9
                  Originally posted by SurvivorSean
                  If I had the money I would hire the programmers myself and start a company.
                  Q: How do you make a small fortune in the Personal Computer - Railway Simulation industry?

                  A: Start with a large fortune!


                  I think the OR team doing this as volunteers and keeping the program pseudo open source is the best way to go. Kind of like Linux for train simming. Only thing I'd suggest is they add a paypal link for people to give them a few donations to at least help subsidize the bandwidth and server storage costs. With many hard-core trainsim volunteers involved and without any money (and the resulting powers from above) this is the best way to create a superior simulation program.

                  Comment

                  • conrail1
                    Senior Member
                    • Jun 2010
                    • 786
                    • NY

                    #10
                    Originally posted by SurvivorSean
                    The funny thing is if Microsoft did further research they would probably understand that if they got it right and spent the amount of time to really do a good job and bring out MSTS2 they probably could of sold it for 99.99 a copy.

                    That’s just it Microsoft did do their “research” and XBOX 360’s future won. With the majority of gamers abandoning the PC completely many of us predicted that MSTS2 wouldn’t see the light of day long before the announcement was made.

                    With MSTS2 using FSX’s game engine it was pretty much doomed from the start anyway. This combined with very little real world railroad background from any of the development staff; it didn’t really show signs of being any different from the rest of train games we’ve seen before.
                    Asus Maximus VI Extreme Intel Z87, Intel Core i7-4770K, Corsair Dominator Platinum 16GB PC3-19200 (2400MHz), EVGA GeForce GTX TITAN SuperClocked, Samsung 840 Pro Series 128GB SSD (Windows 8.1 Pro 64)/Samsung 840 Pro Series 256GB SSD (Programs), CaseLabs Merlin SM8 with Corsair AX1200 PSU

                    Comment

                    • sniper297
                      9500 Post Veteran
                      • Nov 1999
                      • 9545
                      • Union Grove, Wisconsin, USA.
                      • General

                      #11
                      "As we’ve seen in the past with every other train game, “built by railroad enthusiasts” just doesn’t cut it"

                      Problem there is localization, MSTS was originally developed by KUJU - the combined staff at KUJU knew everything there was to know about British commuter railways, so that's what MSTS (and the new birdwatching simulator) was designed for. That's why the front coupler never worked, and why the AI pathing was so screwy, it was all designed for driving passenger trains from station to station on multi track mainlines without ever using the couplers. US freight ops and switching were a last minute afterthought and they didn't know anything about it.
                      sigpic

                      Comment

                      • rdamurphy
                        8000 Post Veteran
                        • Nov 1999
                        • 8429
                        • Where Standard Gauge is 3 feet between the rails.

                        #12
                        You can't blame them for that. As much as has been said about Kuju's programmers, etc, Microsoft also wanted the game built on time and under budget. And of course, we bought it. And still use it 10 years later.

                        I challenge ANYONE to find a game that came out the same year as MSTS that still enjoys such a following today...

                        So, we had to fix a bunch of stuff ourselves, around the world. Isn't that part of the genius of the game? That we could?

                        Seriously...

                        Robert
                        sigpic

                        Comment

                        • SurvivorSean
                          3000 Post Veteran
                          • Apr 2001
                          • 3460
                          • Welland, Ontario, Canada.
                          • General

                          #13
                          You know something for a model railroader (I just like running them anyways) I'm a very budgeted Model Railroader. I never had a huge collection and is also a reason why I joined clubs as an alternative. Pay your dues and run other people's stuff.

                          So if I had the money I'd gladly donate, but when it comes to lack of cash I do what many others do and that is get what they can get for free. Open source despite the massive volunteer effort on the part of say Open Office isn't because it's better than MS Office but it's free in my opinion. I know there are some differences but that is how I see it for my uses.

                          Open Rails is a different beast because it seeks to be the simulator without a budget attached to it. Unfortunately as I type here unemployed developing this doesn't pay the bills. Nor does the team have the resources as if they were working on a project for a software company working 40 plus hours a week and coming home happy they know they're doing something they love and it pays the bills.

                          I like Open Rails and think it's a great concept. But it's going to be a slow climb, and I think from what I've seen thus far it's looking great and I hope it's enjoyable for both those that try it out and the people behind it. For me since my programming skills are a little ancient I'm concentrating more on what I know I can do now, and with RTS I'm doing just that.

                          But being unemployed I can't help to think that if I only could make some money at this like others have done in other ventures I can really make a dent. But it's safe to say it's a risky venture considering how many man hours and resources a project like this requires.

                          Still despite my lack of income I know I'd save and pay a triple digit figure for the ultimate Train Simulator. But I must admit when asked for donations I'd offer little or none. Call it cheap but that's just the way it is - and I'm being honest about it.

                          Thanks

                          Sean
                          Discord https://discord.gg/zfxTBbnu3K
                          YouTube https://www.youtube.com/c/SeanMurrellRTS
                          sigpic

                          Comment

                          • muskokaandtahoe
                            3000 Post Veteran

                            • Apr 2003
                            • 3003
                            • Silicon Valley, CA, USA.
                            • ORTS

                            #14
                            Originally posted by conrail1
                            That’s just it Microsoft did do their “research” and XBOX 360’s future won. With the majority of gamers abandoning the PC completely many of us predicted that MSTS2 wouldn’t see the light of day long before the announcement was made.
                            FWIW Open Rails is being built w/ the .XNA framework. That means w/ a few changes it would run on the XBox. AFAIK nobody on the team has expressed an interest in providing for that so don't hold your breath.

                            For myself, I'm going to try out one of the newer Radeon video cards... I understand it will cable into a big screen TV... and as I happen to own a big screen TV I'm going to be sure to see how OR's looks on it.
                            Dave Nelson
                            sigpic
                            Seldom visiting, posting less often that that.

                            Comment

                            • rdamurphy
                              8000 Post Veteran
                              • Nov 1999
                              • 8429
                              • Where Standard Gauge is 3 feet between the rails.

                              #15
                              Now THAT would be awesome! Multimonitor support, like in FS would be good, too, imagine being able to view multiple cameras simutaneously. I've done it with three monitors in MSFS and I've seen it done with up to 9. That was at a computer store, but it was truly awesome. They had the panel on the bottom center, center console on the bottom right, front window in the center, and window views on the sides and the upper console on the top. I played with it for about 45 minutes, drew quite a crowd, since usually it ran in a "demo" mode because they didn't have anyone who worked there that could fly...

                              Robert
                              sigpic

                              Comment

                              Working...