[net.railroad] Computerized railroad

heliotis (11/09/82)

Is there anyone out there  who has had some experiences with using a
small computer to "run" a toy train set (anything from logic-controlled
switches to route-finding algorithms and planners)?  I have never
heard of anyone doing this (at the toy level), but I figure someone
must have tried.

				Jim Heliotis
				(seismo/allegra)!rochester!heliotis

lee (11/10/82)

I am sure that there must have been several computer run railroads.
The one that comes to mind immediately was at Stanford's Institute
for Mathematical Studies in the Social Sciences.  It was used a part of
a computer education project.

-lee

lou@sri-unix (11/11/82)

I seem to recall that when the LSI-11 was introduced, DEC had an
elaborate, but portable, electric train layout controlled by the
microcomputer.  They brought it to all the appropriate computer
shows.  This would have been in '74 or thereabouts.

			Louis Warshawsky
			General Instrument R&D

jhh (11/11/82)

References: rocheste.212

While at U. of Michigan, I took a course whose last project
required writing a program to control a train on an N-gauge
railroad.  Each block could be powered separately via a D/A
arrangement.  At the entrance to each block were photocells
under the track which detected the darkness of a train and
triggered an interrupt.  Each switch was also under software
control.  Several `throttle controls' were available which
had a switch and a potentiometer.  The pot was fed into
an A/D from which the speed was to be controlled.  Whenever
the pot or switch changed positions, an interrupt was generated.

>From this hardware, the project was to get the train to
respond to the pot for speed/direction control, and the
switch was used for controlling the position of the next
track switch.  Whenever the train left a block, that track
was to be powered down.  Any switch approached from the top
of the Y had to be moved to the correct direction.

For extra credit, multiple trains could be controlled.  This
required collision detection and avoidance plus
dermining which photocell interrupt was associated with which
train.  Other extra credit problems included route selection
and speed control on curves.

There were several problems with this railroad.  One was that
if a light was burned out, or the lights were turned out,
no photocell detection could be done.  Another problem was
that if a second photocell interrupt occurred before
the first one was processed, the second was lost.  Another
less severe problem was that the compiler used was very
inefficient and filled the 24K very quickly.  The language
used was CRASH (unknown acronym), under the OSWIT (Operating
System WIth Trains).

			John Haller

goutal (11/16/82)

I did, at UMASS in 1976.
We had a 4'x8' layout that looked sorta like the olympic games symbol,
inscribed on an oval.  We had a series of projects (for Real-Time
Programming course) of increasing difficulty.
Final project was to run as many as four trains (engines, actually,
-- "quantum trains") through a complete circuit of this thing simul-
taneously, with provision for all sorts of GHA (Giant Hand Action),
such as tampering with switches, physically stopping (like derailing)
various trains, false triggering of train sensors (reed switches under
track).  My partner and I had a sort of state-machine approach to this,
wherein each train responded to various levels of various disaster
by various sequences of action, ranging from simply slowing down
to hitting the brakes to rerouting (to cope with possible brake failure).
Worked pretty well, as I recall.
-- Kenn (decvax!)goutal

wagner (11/18/82)

I think waterloo teaches a course where controlling model trains is the
end result.  I think it is a control systems course.  Anyone in waterloo
listening?
Michael Wagner, UTCS

dmmartindale (11/19/82)

When the original query about computer-controlled model trains was
posted, I responded to the requester by mail, as did others at Waterloo.
There is a course in real-time software which uses a model railroad
(now in its 2nd or 3rd incarnation) as an example of a control problem which
is close to the "real world".
Also, trains are fun to play with.

	Dave Martindale