RTS is back in active development

[SurvivorSean]

Well I don't want to be the boy who cried wolf, because I don't like wolves unless they are cute puppies.

But thankfully the lack of brain power that has let me down in recent coding adventures is now happy to see the AI era is among us. I've been using Grok 4.1 now and it's predecessors for a few months. For the most part anything you can think of. But it wasn't until very recently I have built full applications starting at work with a shipment system, then a simple game of Uno earlier this weekend.

What I'm showing you here I'm calling alpha 0.0001 and so far Grok has been doing well. I've had issues with Grok on more complicated tasks especially when you start getting into 3D and using other sources and programs. But it's very good at writing web based programs. Here is a quick 15 minute video (sorry no sound have to fix my mic) that shows off where RTS is right now.

https://youtu.be/tNnv38UfKOM

So where I'm at right now is I originally wanted to automatically import all of OpenRailroadData but as I said Grok can get confused when you bring in other sources and it did. So I did something I learned from the FYM days and I simply built my own track in an editor. What you are going to see here may look like MS Paint. But I assure you the map is the entire world (just happens to be zoomed in right now to Niagara, Canada between Lake Erie and Ontario).

The system is working perfect at building track nodes and I was able to build the 8 track CN Port Robinson Yard in 15 minutes. There is a similar concept used for building tracks in FYM Freight Yard Manager except RTS is not about just the yards. Instead of using maps people create with just yard images in the background you have access to the latest real world data from the web. You can plot a track anywhere in the world over top existing tracks. What is the point of that, well it goes into a JSON file and not RTS has data to know where the tracks are similar once again to what FYM does.

What FYM does is for free allows people to access the program but you must connect to servers run by the developer. I will be free in an HTML code that anyonce can view, and if they want use if for their own ideas. After all Grok developed it for me and I'm going to share it with the entire world.

What FYM does not do is allow real time AI networks of trains that will AI switch to whatever settings you have (I'm not there yet but its a long way down the pipeline). Now that I have track next will come the cars, engines, and switches where those nodes interact with each other. Signal systems will be developed likely automated as well. At the very least you will be able to run trains anywhere in the world and switch industries. That shouldn't be too far off at the rate I'm going.

Anyways I'm excited about this one and I hope you are as well.

Thanks

Sean

[SurvivorSean]

I've opened RTS to beta for anyone to try out now. The program and data file I have of the track database so far is on my Discord. If you have issues reaching my Discord at the signature link please advise and I will update it and let everyone know here.

I'm open to all comments praise, criticism, ideas, you name it. Hit me with what you got. This is going worldwide so yes even the many here outside of North America will be able to create your lines and networks.

Your input is greatly appreciated.

Thanks

Sean

[SurvivorSean]

Latest version available on Discord is RTS 0.003 which allows worldwide colaboration of rail lines. I was working on getting the first engines online but have run out of time for today. I'll see what I can do after work tomorrow.

Thanks

Sean

[SurvivorSean]

Hello:

CN C-40-8 on the Chemical Lead. RTS 0.0004 is now available. Allows you to place custom PNG image files and flip them. Alligns to the the center of 2 track nodes that you choose. This one took over 30 attempts with Grok to get it as close to perfect as a 2D sim will allow. The engine is courtesy of SMRY(Scott). I'm curious how things look at different latitudes since this is worldwide. I've addressed that with Grok so hopefully we are on the right track pardon the pun

I could use all the beta testers and opinoins as I can get, it is very much appreciated.

Thanks

Sean

image.png

[SurvivorSean]

Last night was a tough night to get 0.004 out. Tonight it took me 15 minutes with Grok and 0.005 is out. Now granted it's still not a train simulator, but I can't believe what it just was able to do with all those tracks I created in Port Robinson. This version is NOT out just yet because it's actually got some comical stuff happening like some weird ghost horn, and a buzzer for a bell. But the controls are not too shabby. Apparently it doesn't have indy or air breaks but I really don't care as I'm simulating the railroad and don't want it to be just another train simulator. Check it out it's a small video. The video contains the receipts I'm on to something now. Trust me I was deflated last night, now I'm way up again.

https://youtu.be/1RtEx5fWalM

[SurvivorSean]

There was a few bugs but I'm now at version 0.005.8 and is up at Discord. Basically you can build a track anywhere in the world, pick the engine of your choosing and drive for as many miles forwards or backwards as you want. I haven't got to switches which will be the next task before I start introducting more advanced topics.

It's kind of sad to see little to no interest here, but I guess because it's 2D what should I expect I suppose. Not to me people seem to share my visions of a train simulator anyways. But since I'm only capable of using AI I had to settle for what it could do.

In fact I was ready to give up again, but I didn't. Instead Grok couldn't solve an issue so I turned to Claude and though it struggled to it's debuging techniques were superior. Grok continues to give me the most free time available where as Claude is very limited for how many messages you can generate per session. One of these days I might try to push Claude hard right from the start to see what it can build and find out if Grok can take over from there. But for now I'll keep grinding on.

I'll probably not post here as much since I seem to be talking to myself at this point with little or no interest in this.

Hope you enjoyed the story of AI and programming at least. I'll continue on with my YouTube and Discord for the time being.

Thanks

Sean

[Leaf85]

Sorry Sean, just not the type of train sim that interests me. I do wish you the best in building it. I was hoping you were back developing RTS for Run 8, but I've a feeling that's run it's course. All the best!

[SurvivorSean]

No I understand and trust me I was trying to build a 3D train simulator at first. This should hopefully be an alternative for the FYM crowd, or simply be something that Scott and I like to run. I kept it worldwide so those outside of North America don't bitch about having to run another North American simulator.

This is to be a full simulator and also act as a proof of concept of what is possible in a 3D simulator because I personally think we can do better than what is out there right now. I just don't have the tools, experience, and trust to build a 3D simulator at this time.

But I do know others are developing new 3d simulators right now.

Thanks

Sean

[SurvivorSean]

Ran into a roadblock late last year, and took a break for a while. The 2D simulator because it wasn't physics based created it's own logic problems.

But last night I learned real quick that AI coding is not as peachy as it sounds (at least not yet). There is a lot of hype out there and it takes someone with real coding skills. So despite trying to solve it's own issues, it just gets you into another rabbit hole.

Thanks

Sean

[SurvivorSean]

I started again from the beginning this time with a clearer vision, and much more AI experience with some programs both personal and work already under my belt.

I'm not going to say too much more about it because I don't want to be the guy that cry's wolf. But I do apprecaite all the support I can get emotionally (I don't need money, and I'm not doing this for money).

I'm doing this for me first and foremost, and two for the community.

Regardless of what the description says this is my Disocrd invite to anyone who wants to know the latest information. Nothing to show right now but plenty of data on what I'm working on. If someone steals this info go for it, I dare you to beat me to it.

Anyways here is my official discord invite open to anyone.

https://discord.gg/Tj54QHnPuP

If someone can remind me how to update my signature again or if an admin can kindly edit it I'd appreciated it. I can't seem to recall where this is done on the site

Again thanks to all for the support.

Sean

[Leaf85]

Good luck with the new development direction, Sean. Honestly I think a lot of corporations are finding AI not as advertised...at least not yet anyway. Cheers!

[SurvivorSean]

You know I had difficulty with AI at first but my experience with ChatGPT+ has been amazing so far for RTS. I am on version 0.011 and have portals with inbound and outbound trains, switching, and all of Niagara rails plotted in full 2D.

This is with 6 days spending a few hours a day. But the thing is I haven't played anything else, I barely watch YouTube. Development has been absolutely awesome. Probably the biggest accomplishment in my life besides my son. Way past the point of no return now. The whole cry wolf because some wall shows up is over. I just crashed through that wall this week.

Thanks

Sean

[SurvivorSean]

Well for the time being I have transfered ownership of the RTS code to SMRY Scott. AI is amazing and in fact I stumbled onto something I'll post in another subject. Essentially what ended up happening is RTS works like it was desgined. Scott is already adding more car types and consist editors which is an attribute to AI since Scott doesn't have the programming background but has plenty of assets he wants to feature.

The reason why I stepeed out of it for the time being is I found Niagara despite knowing the area well, I don't have the full logistics down of every industry and what ships where and how often. The fact that it's hard for me to get around like I used to was that last straw, and I passed over to Scott.

I have rebranded to https://www.youtube.com/@SeanMurrellStudio and I'm now using the lates ChatGPT 6 Astra which as expected is much more powerful and fast. More on that in another thread and how it relates to train simulation.

Thanks

Sean

[SurvivorSean]

Expedition Simulator worked well but I found it was taking too long to get to the trains. So I took what I learned from it and decided to apply it to a new project. I've already developed a proof of concept as I'm using ChatGPT for coding. The flow works well for Unity as the game engine. The new project that I'm on is called RTS:Switching District and I've attached a concept report that I finished tonight with assistance from ChatGPT as we spent a few hours this evening going over the scope of the project.

I'm open to all suggestions. It would be great to get some people interested, but unfortunately all I can do is what I'm capable of. AI is starting to get much better and it's breathed new life into development this past year.

__________________________________________________ _________ SEAN MURRELL STUDIO / DESIGN CONCEPT RTS: Switching District

Concept Design Document • Discussion Draft v0.1 • 24 October 2026

A procedural, autonomous railroad operations simulator focused on meaningful switching work within a larger fictional railroad network.

1. Vision and scope

RTS: Switching District is a modern, fictional North American-style railroad operations simulator. The focus is operating an established railroad: classification, local switching, train movements, dispatching, and the interaction between local work and the wider network. It is not a railroad-building tycoon, financial management game, or primarily a locomotive-driving simulator.

The simulation runs autonomously. A player may observe, assume a role (such as conductor, yardmaster, dispatcher, or switcher operator), and return that role to AI. The aim is meaningful railroad work without mandatory micromanagement.

2. The simulated world

Generate one fictional, Earth-like continent or continental region—not a whole planet. Broad coastlines, major lakes, rivers, mountains, plains, forests, resources, ports, and settlement patterns provide context. Geography should influence economic activity and transportation corridors, but detailed external terrain is unnecessary.

Resources may be abundant, scarce, or absent. Ports connect to abstract overseas supply and demand; other railroads provide destinations beyond the primary railroad's reach. No ocean voyages or foreign continents need to be simulated.

Generate a contemporary, established economy rather than simulating centuries of development. Broad regional production and consumption are sufficient outside the playable area.

3. Railroad network and organization

The primary railroad is an abstract network of major locations and connections, with distances, approximate travel times, capacity, grades, route restrictions, terminals, and interchange points. It may be a large Class I-style system, a regional railroad, or a smaller operation; railroad classes are scale presets rather than a need for detailed financial accounting.

Divisions and subdivisions describe railroad organization and routes. Switching districts describe operational service responsibilities and do not have to align with subdivision boundaries. Multiple districts may use one subdivision; one district may work across several subdivisions. Crew districts and relief points are separate from subdivision endpoints.

Other railroads can overlap geographically, connect by interchange, or operate under trackage rights. Track ownership, train ownership, and dispatching authority are distinct.

4. Switching districts and playable switching regions

A switching district consists of yards, customers, and assignments served as a coherent operating responsibility. Districts are sized by work and travel time, not a fixed radius. Multiple crews and supporting yards are possible.

A playable 'switching region' (working RTS term, not asserted to be an official railroad designation) can encompass one district or several closely interacting districts. Include complete routes required to perform their switching work, even when 80–100 km or more at 1:1 scale. The region can follow irregular corridors rather than a rectangular map or subdivision limits.

Neighbouring districts may share tracks, junctions, and transfer movements without having identical industry responsibilities. A nearby district is included when its local operations meaningfully intertwine; ordinary distant road-train connections remain abstract.

The player can work any relevant yard, industry, or train within the selected region. Uncontrolled operations continue autonomously.

5. Freight economy and car flows

Commodity supply chains explain freight movement. Broad geographic resources, producers, processors, consumers, ports, and interchange gateways create inbound, outbound, and through traffic. External regions need not enumerate every factory; the playable region expands relevant demand into actual industries and individual car movements.

Initially emphasize offline-to-online, online-to-offline, and offline-to-offline (bridge) movements. Online-to-online shipments are not an initial priority.

Freight car type, origin, destination, routing, train assignment, and classification needs should be consistent with the commodity and railroad network. Industry demand varies by day without requiring the industry landscape to evolve over decades.

6. Train services and traffic

Bridge trains can traverse the region without stopping. Unit coal, intermodal, autorack, passenger, and other services should appear when justified by the larger network—even if the playable region has no corresponding facility. A facility affects whether a train stops, not whether that train can pass through.

Road manifests may exchange blocks at a yard and continue, or terminate for classification when volume warrants. Smaller yards may be served by one or two designated trains; major terminals may originate, terminate, or work many trains. Road-train patterns and yard roles arise from demand and network topology.

Use recurring services and plausible call times with variability and delays. Passenger service is more timetable-driven and exists only on suitable routes. Train trips and crew duties are separate; virtual crew changes and jurisdiction-specific hours-of-service limits must be considered. Ten hours may be a planning target, not a universal legal rule.

Virtual route grades and restrictions may explain locomotive consists and distributed power, even when the playable district itself is relatively flat.

7. Detailed railroad simulation

Generate actual 1:1 track geometry and plausible terrain only for the selected region. Include yards, ladders, crossovers, sidings, junctions, industrial tracks, grades, bridges, signals, and appropriate surrounding scenery. Infrastructure is justified by traffic, terrain, and historical constraints—not idealized layouts.

Support single-, double-, and mixed-track routes, bidirectional movements, and appropriate CTC, ABS, track-warrant, or mixed operating methods. Distinguish movement authority from signal indication.

Use a data-driven track graph separate from Unity 3D rendering. Train operation begins with simplified but meaningful physics: length, mass, acceleration, braking, grades, and coupling. More detailed train handling can be added later.

8. Player roles and autonomous operation

Initial hands-on interaction is conductor-like: queue commands to proceed, stop, couple, uncouple, shove, wait, or resume, while an AI engineer handles locomotive controls. The command queue is a working concept, not a finalized interface.

AI and player use the same movement-authority, routing, and switching systems. Players can take over and relinquish trains, yards, industrial switching, and dispatching without halting other operations. Future direct driving is optional.

9. Procedural generation and persistence

Proposed conceptual order: broad geography → resources and settlements → economic activity and freight demand → primary and connecting railroad networks → divisions, terminals, subdivisions, switching districts, and train services → selection of an operationally coherent switching region → detailed 1:1 generation.

These stages may iterate: railroad access influences industry, and industry justifies railroads. The abstract network records approximate workloads, yard roles, industry counts, route connections, and traffic without creating detailed track for every district.

Offer guided criteria and a surprise/random option. Show a lightweight operating profile of candidate regions before the expensive detailed generation. Once accepted, the region is saved persistently; reopening it does not regenerate the tracks. Deterministic seeds and multiple saves are desirable.

10. Technical direction and development sequence

Target Unity 6.3 LTS and the current modest PC initially (i7-3770K, 8 GB RAM, GTX 1650 SUPER 4 GB). An earlier small autonomous warehouse scene established a basic Unity workflow, not proof of large-scale performance. Large regions and dense traffic will require streaming, level-of-detail, and efficient simulation.

Suggested build order differs from the finished world's generation order: (1) a small manually defined operating test railroad with autonomous switching, track graph, car routing, and movement commands; (2) procedural detailed yard/district generation; (3) abstract continent, economy, and full railroad network; (4) integrated selection and persistent operations.

Use data-driven definitions for commodities, industries, rolling stock, locomotive properties, and generation parameters. Preserve the possibility of editors and mods; no editor or full modding interface is required for the first prototype.

11. Open design questions for review

How should local assignment planning estimate work time, switching difficulty, crew availability, and yard capacity?

How should inter-yard transfers distinguish tightly coupled neighbouring districts from distant road-freight connections?

What minimum network and economic information is necessary to generate believable train consists and service patterns?

How should a detailed region transition to abstract operations at each boundary without losing train identity or timing?

What are the practical performance limits for 1:1 regions and busy corridors on the target PC?

Which operating practices vary by railroad, region, era, and jurisdiction, and therefore belong in configurable rules?

12. Status

Concept draft for discussion, dated 24 October 2026 (Eastern time). This document records design intentions, not finalized implementation commitments. Real-world examples—including Niagara/Port Robinson, Hamilton, Aldershot, Oakville, and major western terminals—are inspirations, not promises of prototype-accurate maps. Feedback and corrections are welcome. RTS: Switching District • Concept draft for review

Thanks

Sean