# What I planned to do
After passing a looooot of time documenting the state of the art on the
hardware side, and evaluating the few detection methods that are
available and not in the state of the art: I've finally decided what to
develop.
The majority of methods that I have documented only detect the passing of
a train or a block occupation. The only few that directly measure the
precise position are either impracticable in some situation or completely
vanished.
So, I decided to use 2 consecutive IR detectors, working by reflection
against the train, to measure speed. By using the raw value of the
detectors, I also hope to detect where begin and end each vehicle and
where the coupling are. Or even each vehicle signature. Then, using a
modified JMRI, calculate the exact position of the train and where each
vehicles and coupling are.
Of course, this only work if trains maintain their speed after the
detectors, or only stop and restart. But, I could also use the speed
values measured by the detectors to build a speed table of each
locomotive. And then use these tables to estimate new speed in case of
speed change.
After having the precise position of the trains and wagons, I planned to
modify JMRI add what miss to manage train composition and de-composition,
and wagon distribution, from a high level interface with abstraction of
the moving management needed for each maneuver.
It might be a little late for this, but can I suggest another way of detecting position that might both (1) meet your goals and (2) be fun to work on?
Put a video camera above the layout, and use visual pattern recognition to find the locations of the individual pieces.
People have played with this a little, but as far as I know nobody has made it work well. The technology wasn¡¯t up to it. But maybe it is now?
Bob
¡ª
Bob Jacobsen
rgj1927@...
Bob Jacobsen