The Great Model Railway Giveaway - Part 2

[hicksan]

8 hours ago, woodenhead said:

My eyes did read that last sentence as "Strategically positioned pensioners etc would also help here".

 

...

 

A question to @hicksan will the wagons to be shunted be required to be positioned over specific sensors at set spots on the layout or is there some other method of identifying all the wagons spotted at the start without relying on something under the track?   I guess I am considering how an intelligently automated inglenook with standard wagon lengths might evolve to a layout less uniform with both short and longer wheelbase wagons in the mix that may not fit positioned sensors.  Or is it simply a case of a lot of sensors capable of deducing the centre of a wagon and therefore it's length?

 

Secondly, might you consider making your couplings servo or magnetically assisted in some way so that they are not reliant on under track magnets to operate, again so your automation rests more in the real world that a wagon may not always be left in the exact same position albeit in the right order.

 

MERG has a good number of "strategically positioned pensioners" who I'm sure will be keen to help me solve this.

 

The questions of identifying which wagon is where and how to uncouple particular wagons without too many unnecessary extra movements are both at the heart of solving the practical part of this problem. Each has several possible solutions and I am not going to commit to any one of them yet, but it will be a fun process testing each till I find what works best.

 

However my initial thought is that a single RFID sensor at the mouth of the points ought to be able to identify anything rolling over it. From that it will be able to retrieve data as to the length of each vehicle. If it knows the route set and the direction of movement (which it ought to since it will be driving) then it will be able to figure out exactly what is where and which have moved. The 'setup' process would require the loco to move the wagons into position, or could be entered manually via the control interface (probably a touch screen).

 

RFID is a subject all by itself, where range and reliability are key. MERG members have been experimenting with the RC522 receivers and have already made a CBUS interface for it. Other groups, such as McKinley have had success with the Eccel Pepper-C1.

 

With regard to couplings; again there are numerous options for "delayed uncoupling". I hesitate to fit electronic uncouplers in every wagon on grounds of cost. Personally I have used both Spratt & Winkle (mk2) and Kadee couplers and prefer the former (at least of wagons). I will probably then have a single uncoupling magnet near to the RFID sensor, and probably mounted on a servo so as not to cause accidental uncoupling of slow-moving trains. I figure that if the system knows the exact position of each wagon as it passes over the RFID sensor, combined with the length of the wagon, the speed and direction of the train and its stopping distance; then it ought to be able to figure out exactly where to stop the loco so as to position the correct couplings over the magnet before activating the servo.

 

But all of this is just conjecture at this stage. I look forward to running some experiments on my test track and sharing  progress (or failure) along the way.

 

These are just the practical problems. I still have to figure out how best to solve an Inglenook with the fewest moves for any given wagon arrangement. My thought is - once I can reliably move wagons, collect, drop off and identify each - to code up several different approaches to run against the initial setup as simulations, before doing any actual movements, and then select whichever solution comes up as most efficient. It's going to be an interesting journey.

[woodenhead]

@hicksan As I read through that I have to admit I am beginning to imagine a hump marshalling yard scenario as a scaled up version.

 

RFID sensors detecting wagons passing over an uncoupling magnet which fires at the appropriate cut point and then perhaps some electromagnetic solution to keep the wagons moving along at a set pace - not too fast (which is what happens in any traditionally modelled hump) or to slow as to stop after a few centimetres.

 

[JSModels]

Reading your plans for automating the layout immediately struck a chord with me - when I first built my shunting puzzle I wondered if it would be possible to automate one. Having exactly zero skills or knowledge in that area however immediately made a theoretical idea only!

 

I did think it would be really cool if it could be done, and I did wonder if it had already been done somewhere by someone, but maybe it never has. Until now...

 

(The fact that Phil used my buildings on the layout is an added bonus!!!)

 

Jonathan

JSModels 

Edited March 21, 2023 by JSModels

[hicksan]

1 hour ago, woodenhead said:

@hicksan As I read through that I have to admit I am beginning to imagine a hump marshalling yard scenario as a scaled up version.

 

RFID sensors detecting wagons passing over an uncoupling magnet which fires at the appropriate cut point and then perhaps some electromagnetic solution to keep the wagons moving along at a set pace - not too fast (which is what happens in any traditionally modelled hump) or to slow as to stop after a few centimetres.

 

This has in fact been done, and in the 1960s to boot.

You can find the video on Facebook under the title "1962: Machines Like Men"

The National Museum of Computing at Bletchley Park is currently building a recreation of this. I took a look at the original computer code once and it is brilliantly simple. One could do it with an Arduino Uno these days.

Yes it uses logic, so is proper automation, and absolutely it is inspirational but the loco movements are quite basic. It isn't something that can be generalized onto any layout as I am intending. For me the Inglenook is merely proof of concept, not an end in itself.

 

 

[hicksan]

14 minutes ago, JSModels said:

Reading your plans for automating the layout immediately struck a chord with me - when I first built my shunting puzzle I wondered if it would be possible to automate one. Having exactly zero skills or knowledge in that area however immediately made a theoretical idea only!

 

I did wonder if it had already been done somewhere by someone, but maybe it never has. Until now...

 

(The fact that Phil used my buildings on the layout is an added bonus!!!)

 

Jonathan

JSModels 

I am approaching the code as nested scripts, with the base level made up of a repertoire of simple movements and then calling these to execute the sequence once the logic part has figured out what to do. In theory this could be generalized to any layout. The long term plan is to automate stations or fiddle yards within larger layouts - along  the lines of the famous Automatic Crispin only able to drive trains and shunt them.

 

The buildings are lovely, btw.

[JamFjord]

I'm really enjoying this discussion - hopefully it continues here on RMweb and also encourages some participants to join MERG.

 

22 hours ago, hicksan said:

You can find the video on Facebook under the title "1962: Machines Like Men"

 

For those like me not on Facebook, here's the video on YouTube - an interesting watch. Very much of its time, but it was worth reminding myself that this was broadcast over 60 years ago!

 

 

22 hours ago, hicksan said:

...but the loco movements are quite basic. It isn't something that can be generalized onto any layout as I am intending. For me the Inglenook is merely proof of concept, not an end in itself.

 

This really is a fascinating idea, Andrew - I'm glad I caught the E101 chat on Monday and I'll be following progress on the automation of this layout with great interest. Coming up with a system that will be adaptable or configurable to any layout is quite the challenge to take on. I'm with Huw in that a lot of this is beyond me at present, but automation is definitely something I'd like to start experimenting with at some point. In the meantime, though, I probably have enough on my plate bringing my own layout plans to fruition in time for Doncaster next Feb... I'll be using MERG tech to help me do that, but not pushing the envelope like yourself.