|
Hello,
?
I have a few questions.
?
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
?
If yes:
- With which precision ?
- Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
?
Do you know of other software than JMRI who can do that ?
?
?
For my diploma work for engineering, I need to develop a system that can measure, or deduce, the precise location of a train, with a precision of 2-3 cm.
Then, using this location, the system need to automate the maneuvers, including uncoupling and train formation.
?
But before developing this system, I need to document what's already exist in the field of model making. And then found a new way of doing it.
I'm only allowed to develop something that nobody have already done.
?
For the detectors, it's seems that model makers have already done every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
?
For positon deduced from other data, I have seen that JMRI already have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
?
Best regards.
?
PS: If my diploma work can be done, I plan to release the result software and hardware as free (as in freedom) software and hardware.
?????? And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as contribution to JMRI.
?
[1] https://www.jmri.org/help/en/package/jmri/jmrit/logix/Tracker.shtml
|
Warrants use block and path lengths with train speed to calculate position. It also has the capability to adjust the warrant to reflect actual running.
?
--
H.O. Australia (Layout in Progress)
Digikeijs DR5000 LocoNet
JMRI v5.8 DecoderPro/Warrants/CPE/SML/LogixNG
Windows 10
|
From what I I know about JMRI, it will do the things you ask for, however there a very steep learning curve, in my opinion and you need to understand programing to get the best results.
A better solution, in my humble opinion (still with a learning curve, but much less than JMRI) is to use TrainController (Gold Version), apart form automation it is very easy to draw your layout in TrainController. Another advantage to TrainController is that there are very good YouTube videos available from several authors that will take you through the process of learning how to use TrainContoller.
?I understand that the price of TrainController Gold Version at approximately US$800+ (less costly versions of TrainController are available at lower cost and functionality), will be off-putting for some people. However if you have a DCC based train layout your investment in DCC technology will probably be in the several thousands of $. For example the price of x2 well equipped high end ?loco’s with sound can cost more than TrainController.Gold
I am sure that others will have varying levels of viewpoint than mine.
toggle quoted message
Show quoted text
On Aug 7, 2024, at 8:51?AM, Sébastien Gendre via groups.io <seb@...> wrote:
Hello,
?
I have a few questions.
?
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
?
If yes:
- With which precision ?
- Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
?
Do you know of other software than JMRI who can do that ?
?
?
For my diploma work for engineering, I need to develop a system that can measure, or deduce, the precise location of a train, with a precision of 2-3 cm.
Then, using this location, the system need to automate the maneuvers, including uncoupling and train formation.
?
But before developing this system, I need to document what's already exist in the field of model making. And then found a new way of doing it.
I'm only allowed to develop something that nobody have already done.
?
For the detectors, it's seems that model makers have already done every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
?
For positon deduced from other data, I have seen that JMRI already have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
?
Best regards.
?
PS: If my diploma work can be done, I plan to release the result software and hardware as free (as in freedom) software and hardware.
?????? And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as contribution to JMRI.
?
[1] https://www.jmri.org/help/en/package/jmri/jmrit/logix/Tracker.shtml
|
My first step would be to profile locomotive speed with the JMRI profiler (bottom of the "roster" menu).
That correlates speed step with actual locomotive speed over the entire configured decoder speed range.
So, if you know the throttle speed step as you enter a detected section of track and the length of the section, you can calculate position (d=v/t).
Assuming, of course, that the speed step doesn't change.
Otherwise, you need more sophisticated technology like those radar distance detectors used in robots.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
Be aware, the speed profile DOES NOT include any configured momentum.
So, if you want exact stopping distance, you won't get it if there is deceleration configured.
Likewise, configured acceleration will make startup positioning inaccurate until the locomotive hits the next detected section.
(The natural startup momentum of the loco mechanism contributes to that inaccuracy).
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
D=VT
Oh yes, and since the profile doesn’t have every step mapped, you interpolate the speeds between steps. Theoretically, you could map each step, but practically, well, you’d have to break for lunch.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
厂é产补蝉迟颈别苍, Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ? Yes. I have used the "LayoutEditor-Train Simulator Proof of Concept" to implement sound-follows-loco in VSD. See the attached screenshot. JMRI PanelPro and a LayoutEditor panel is needed. Here is the location calculation: With which precision ? I think with a precision of 2-3 cm. Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ? Partly. Not sure how "uncoupling" could be done. And I only used locomotive info, no train info. Klaus Am 07.08.2024 um 14:51 schrieb Sébastien Gendre: Hello,
I have a few questions.
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
If yes:
* With which precision ?
* Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
Do you know of other software than JMRI who can do that ?
For my diploma work for engineering, I need to develop a system that can measure, or deduce, the precise location of a train, with a precision of 2-3 cm.
Then, using this location, the system need to automate the maneuvers, including uncoupling and train formation.
But before developing this system, I need to document what's already exist in the field of model making. And then found a new way of doing it.
I'm only allowed to develop something that nobody have already done.
For the detectors, it's seems that model makers have already done every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
For positon deduced from other data, I have seen that JMRI already have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
Best regards.
PS: If my diploma work can be done, I plan to release the result software and hardware as free (as in freedom) software and hardware.
And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as contribution to JMRI.
[1]
|
Sorry, here is the attachement.
Klaus
Am 08.08.2024 um 15:44 schrieb Klaus Killinger:
toggle quoted message
Show quoted text
厂é产补蝉迟颈别苍,
> Does JMRI can deduce the precise location of a train from its speed,
it's size and a few object detectors ?
Yes. I have used the "LayoutEditor-Train Simulator Proof of Concept" to implement sound-follows-loco in VSD. See the attached screenshot. JMRI PanelPro and a LayoutEditor panel is needed.
Here is the location calculation:
> With which precision ?
I think with a precision of 2-3 cm.
> Does JMRI can use train precise location to automate the maneuvers,
including uncoupling and train formation ?
Partly. Not sure how "uncoupling" could be done. And I only used locomotive info, no train info.
Klaus
Am 07.08.2024 um 14:51 schrieb Sébastien Gendre:
Hello,
I have a few questions.
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
If yes:
* With which precision ?
* Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
Do you know of other software than JMRI who can do that ?
For my diploma work for engineering, I need to develop a system that can measure, or deduce, the precise location of a train, with a precision of 2-3 cm.
Then, using this location, the system need to automate the maneuvers, including uncoupling and train formation.
But before developing this system, I need to document what's already exist in the field of model making. And then found a new way of doing it.
I'm only allowed to develop something that nobody have already done.
For the detectors, it's seems that model makers have already done every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
For positon deduced from other data, I have seen that JMRI already have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
Best regards.
PS: If my diploma work can be done, I plan to release the result software and hardware as free (as in freedom) software and hardware.
And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as contribution to JMRI.
[1]
|
My one comment about trying to track trains is that these are not high
precision devices. Most have pretty wide variance in speed while they
warm up. I've also seen them stable for a while and then after a
couple of hours running to start changing response again. These are
small motors and usually combinations of plastic and metal gears.
If your programming concept is able to recognize this sort of drifting
about and adapt to it, that would be quite a challenge. But
conceptually possible.
-Ken Cameron, Member JMRI Dev Team
www.jmri.org
www.fingerlakeslivesteamers.org
www.cnymod.com
|
My thoughts still gravitate to Arduino controlled robot radar detectors if your requirements are 2-3 cm.
If you plan on uncoupling with magnets like Kadee, +/- 3 cm might not hack it.
A Kadee uncoupler is only 5cm long with a "sweet spot" maybe +/-2 cm by my experience with well tuned trip pins.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
If you are running automation, you might try the JMRI dispatcher.
Let's say your train approaches an uncoupling position.
As soon as the train enters the detected section (current sense), it will slow down.
To precisely stop that train, a photocell or IR detector will stop the train on a designated spot.
("stopping sensor")
Now, that's only the head end.
Make sure that the coupler is correctly positioned to separate that 40', 50', or X' car.
You may end up with an IR detector at the coupler, count the cars, and stop with logicNG or script.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
Why assume stationary uncouplers? On-board DCC ones provide much more freedom (yes, at much more expense, but we don't know the size and scale of the layout envisioned, nor the budget, nor the number of wagons (that would all have to be fitted with them, and with a function decoder)).
I am using a scaled down version, where rakes of passenger cars are pretty much fixed, but locos can decouple via DCC and run around for recoupling at the other end.
Also, the desire stated in the OP was to build on JMRI, not just use what is there as that would not be an acceptable result.
Wouter
If you are running automation, you might try the JMRI dispatcher.
Let's say your train approaches an uncoupling position.
As soon as the train enters the detected section (current sense), it will slow down.
To precisely stop that train, a photocell or IR detector will stop the train on a designated spot.
("stopping sensor")
Now, that's only the head end.
Make sure that the coupler is correctly positioned to separate that 40', 50', or X' car.
You may end up with an IR detector at the coupler, count the cars, and stop with logicNG or script.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
I think the OPs desire is to work on a new method of sensing train positions with a view to writing a thesis, presumably towards a Masters or PhD (he says diploma, but this seems higher level than that). He seems to want to test out whatever methods he is investigating using models, possibly before attempting to apply it to full size trains which I imagine would be running fully automated. At least that is my reading of his original post.
Why assume stationary uncouplers? On-board DCC ones provide much more freedom (yes, at much more expense, but we don't know the size and scale of the layout envisioned, nor the budget, nor the number of wagons (that would all have to be fitted with them, and with a function decoder)).
I am using a scaled down version, where rakes of passenger cars are pretty much fixed, but locos can decouple via DCC and run around for recoupling at the other end.
Also, the desire stated in the OP was to build on JMRI, not just use what is there as that would not be an acceptable result.
Wouter
If you are running automation, you might try the JMRI dispatcher.
Let's say your train approaches an uncoupling position.
As soon as the train enters the detected section (current sense), it will slow down.
To precisely stop that train, a photocell or IR detector will stop the train on a designated spot.
("stopping sensor")
Now, that's only the head end.
Make sure that the coupler is correctly positioned to separate that 40', 50', or X' car.
You may end up with an IR detector at the coupler, count the cars, and stop with logicNG or script.
--
Ken NYNH&H, Old Colony Division, Cape Cod Branch (1949-1959) Loconet * JMRI 5.9.2 * OSX,Win10,Ubuntu
Blog: ?
|
A warrant is a fixed point on the map that tell which train pass by it??
"Nags via groups.io" <snowy999@...> writes:
toggle quoted message
Show quoted text
Warrants use block and path lengths with train speed to calculate position. It also has the capability to adjust the warrant to reflect actual
running.
JMRI: Warrants Help
JMRI: Warrant Speeds Help
|
I have considered the radar distance detector
But it have some disadvantage: It need to be installed in the axis of a
straight track and have no obstacle.
If the track is the main one of the station, and at the in and out of
the station you also have straight track, you can't install the radar.
And if you do train maneuver and place a wagon between your train and
the radar, it can no longer measure train position.
Now, I am thinking of using 2?light sensors (with IR) to detect train
arriving, measure its speed and signature (amount of IR reflected).
From it's speed and signature, calculate it's length and coupling
position.
And from speed, time and length, calculate its position.
But if JMRI already do the calculation part, it wont be original enough.
And I maybe need to found another idea.
"Ken Heywood via groups.io" <kheywood@...> writes:
toggle quoted message
Show quoted text
My first step would be to profile locomotive speed with the JMRI profiler (bottom of the "roster" menu).
That correlates speed step with actual locomotive speed over the entire configured decoder speed range.
So, if you know the throttle speed step as you enter a detected section of track and the length of the section, you can calculate position (d=v/t).
Assuming, of course, that the speed step doesn't change.
Otherwise, you need more sophisticated technology like those radar distance detectors used in robots.
|
From what I understanding:
With warrants, JMRI can calculate the position of the train after it
entering into a block and using the train speed and block length.
But the position calculated only tell the position of the train as a
point on the track. It did not know where the train begin and end on the
track?? And it did not know the position of the couplings of the train??
"Nags via groups.io" <snowy999@...> writes:
toggle quoted message
Show quoted text
Warrants use block and path lengths with train speed to calculate position. It also has the capability to adjust the warrant to reflect actual
running.
JMRI: Warrants Help
JMRI: Warrant Speeds Help
|
I have difficulties to understand what a "sound-follows-loco in VSD" is.
(Sorry, english is not my main language).
Does the "LayoutEditor-Train Simulator Proof of Concept" is only a
simulation of train??
By position, do you detect/calculate the exact position AND occupation
of the train?? Like where it begin and where it end on the track and where
the couplings are on the train??
How do you precede to detect/calculate the train position??
"Klaus Killinger" <klaus.killinger@...> writes:
toggle quoted message
Show quoted text
厂é产补蝉迟颈别苍,
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
Yes. I have used the "LayoutEditor-Train Simulator Proof of Concept"
to implement sound-follows-loco in VSD. See the attached screenshot.
JMRI PanelPro and a LayoutEditor panel is needed.
Here is the location calculation:
With which precision ? I think with a precision of 2-3 cm.
Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
Partly. Not sure how "uncoupling" could be done. And I only used
locomotive info, no train info.
Klaus
Am 07.08.2024 um 14:51 schrieb Sébastien Gendre:
Hello,
I have a few questions.
Does JMRI can deduce the precise location of a train from its speed,
it's size and a few object detectors ?
If yes:
* With which precision ?
* Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
Do you know of other software than JMRI who can do that ?
For my diploma work for engineering, I need to develop a system that
can measure, or deduce, the precise location of a train, with a
precision of 2-3 cm.
Then, using this location, the system need to automate the
maneuvers, including uncoupling and train formation.
But before developing this system, I need to document what's already
exist in the field of model making. And then found a new way of
doing it.
I'm only allowed to develop something that nobody have already done.
For the detectors, it's seems that model makers have already done
every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
For positon deduced from other data, I have seen that JMRI already
have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
Best regards.
PS: If my diploma work can be done, I plan to release the result
software and hardware as free (as in freedom) software and hardware.
And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as contribution to JMRI.
[1]
|
About the position detection/calculation in JMRI:
Does it JMRI can calculate the position of each vehicle of a train??
Like, for example, telling that a train begin at cm X and end at cm Y, with couplings
at cm A, B, and C??
About TrainController, I've searched but found nothing about train
position other than telling which block is occupied. Do you have more
information??
"Ian Blease" <ianbritmanone@...> writes:
toggle quoted message
Show quoted text
From what I I know about JMRI, it will do the things you ask for, however there a very steep learning curve, in my opinion and you need to
understand programing to get the best results.
A better solution, in my humble opinion (still with a learning curve, but much less than JMRI) is to use TrainController (Gold Version), apart form
automation it is very easy to draw your layout in TrainController.
Another advantage to TrainController is that there are very good YouTube videos available from several authors that will take you through the
process of learning how to use TrainContoller.
I understand that the price of TrainController Gold Version at approximately US$800+ (less costly versions of TrainController are available at
lower cost and functionality), will be off-putting for some people. However if you have a DCC based train layout your investment in DCC
technology will probably be in the several thousands of $. For example the price of x2 well equipped high end loco’s with sound can cost more
than TrainController.Gold
I am sure that others will have varying levels of viewpoint than mine.
Ian Blease
On Aug 7, 2024, at 8:51?AM, Sébastien Gendre via groups.io <seb@...> wrote:
Hello,
I have a few questions.
Does JMRI can deduce the precise location of a train from its speed, it's size and a few object detectors ?
If yes:
* With which precision ?
* Does JMRI can use train precise location to automate the maneuvers, including uncoupling and train formation ?
Do you know of other software than JMRI who can do that ?
For my diploma work for engineering, I need to develop a system that can measure, or deduce, the precise location of a train, with a
precision of 2-3 cm.
Then, using this location, the system need to automate the maneuvers, including uncoupling and train formation.
But before developing this system, I need to document what's already exist in the field of model making. And then found a new way of
doing it.
I'm only allowed to develop something that nobody have already done.
For the detectors, it's seems that model makers have already done every possibilities.
Including using 2 objects detectors to measure speed (and JMRI already have a speedometer).
For positon deduced from other data, I have seen that JMRI already have a Train Tracker [1].
But, as I have only used the Decoder Pro until now, I don't know if the Train Tracker correspond to what I'm asked to develop or not.
Best regards.
PS: If my diploma work can be done, I plan to release the result software and hardware as free (as in freedom) software and hardware.
And all the documents in similar licence. And if it implies modifying JMRI, I would be very happy to send my modifications as
contribution to JMRI.
[1]
|
厂é产补蝉迟颈别苍, Am 06.09.2024 um 14:31 schrieb Sébastien Gendre: I have difficulties to understand what a "sound-follows-loco in VSD" is.
(Sorry, english is not my main language). VSD is the "Virtual Sound Decoder" application in JMRI. VSD provides a way for those who do not have sound-equipped locomotives (or who are not satisfied with the hardware sound decoder options in smaller scales) to enjoy throttle-controlled locomotive sounds from their computer speakers. My goal with VSD was to change the sound position accordingly to the locomotive position. JMRI's 3D audio capability (OpenAL 3D sound based) allows to send the sound to a position on the model railroad layout. I called this "sound-follows-loco". (my native language is german). Does the "LayoutEditor-Train Simulator Proof of Concept" is only a
simulation of train?? Yes, but the implementation of the concept in VSD can drive real model railroad locomotives. By position, do you detect/calculate the exact position AND occupation
of the train?? Like where it begin and where it end on the track and where
the couplings are on the train?? I only calculate the position of the locomotive within a track segment of the layout. The length of the train is not known. How do you precede to detect/calculate the train position?? A detection device is not necessary. But VSD also supports locomotive tracking hardware, which is recommended. VSD assumes that the starting position is the center of a track segment. When the locomotive moves, the position is calculated. VSD knows through the layout the starting and ending point of the track segment, the angle and the distance. JMRI provides the locomotive speed and the traveling direction. Klaus
|
Thank you very much for your answers.
As I was reading the "Warrants" manual page [1], in the section "The
Importance of Block Path Lengths", I have seen that JMRI?detect the
position of a train when it enter a block.
But this section also say that: "All other positional information must
be calculated".
Does that mean that JMRI can calculate the train position inside a block??
[1]
"Klaus Killinger" <klaus.killinger@...> writes:
toggle quoted message
Show quoted text
厂é产补蝉迟颈别苍,
Am 06.09.2024 um 14:31 schrieb Sébastien Gendre:
I have difficulties to understand what a "sound-follows-loco in VSD" is.
(Sorry, english is not my main language). VSD is the "Virtual Sound Decoder" application in JMRI. VSD provides a
way for those who do not have sound-equipped locomotives (or who are
not satisfied with the hardware sound decoder options in smaller
scales) to enjoy throttle-controlled locomotive sounds from their
computer speakers.
My goal with VSD was to change the sound position accordingly to the
locomotive position. JMRI's 3D audio capability (OpenAL 3D sound
based) allows to send the sound to a position on the model railroad
layout. I called this "sound-follows-loco".
(my native language is german).
Does the "LayoutEditor-Train Simulator Proof of Concept" is only a
simulation of train?? Yes, but the implementation of the concept in VSD can drive real model
railroad locomotives.
By position, do you detect/calculate the exact position AND occupation
of the train?? Like where it begin and where it end on the track and where
the couplings are on the train?? I only calculate the position of the locomotive within a track segment
of the layout. The length of the train is not known.
How do you precede to detect/calculate the train position?? A detection device is not necessary. But VSD also supports locomotive
tracking hardware, which is recommended.
VSD assumes that the starting position is the center of a track
segment. When the locomotive moves, the position is calculated. VSD
knows through the layout the starting and ending point of the track
segment, the angle and the distance. JMRI provides the locomotive
speed and the traveling direction.
Klaus
|
Ken Heywood
Ken Cameron