FlyPT Mover

It lags for sure, but in the video, communication was over www, not inside my network.
Considering this, I think it was fast.
What we receive is linear accelerations with or without gravity. So it's not easy to look at it and see if that's good or not.
I think accelerations are precise.
Problem is more on rotations, because sensor sends only the rotation speed.
I can calculate the rotation position, from the difference on linear accelerations (with and without gravity). Knowing the component of gravity in each axis, we can calculate the rotations.

Needs some more work...

And I did not forgot the forward kinematics. I'm working on that also.

It seems it writes to the phone with 100hz.
At the phone graph I don’t see any lag.

Now I want to synchronize the data with the vid in simtools but curiously it doesn’t want to eat it.

@yobuddy ,
It says that import of raw gyro accel telemetry from phyphox file in video ride creator was successful but I can’t see anything.

It seems we have an issue with conversion to 16bit values or something.

This is how phyphox looks like:



Edit:
It seems the renamed .csv file from phyphox has to include a parenthesis:

file_name(100, Tab).cfg

Cool !
I've pushed all sliders up in tools. Now I cant understand why my yaw and heave draws identical curve.

Any ideas?

No, you need to install a new plugin.
Different and more data in new plugin (12 axis)[/QUOTE]

@pmvcda
where I can find a new plugin ? I' m a lite bit lost , Sorry !!
Now I have a new version of simtool 2.4

with this plugin ...
​

Source simtools don't want connect .. missing FlyPt Mover


Thank you !

You need a specific plugin for Mover
It's the dll included in the download.

FlyPT_Mover_InterfacePlugin.dll

Hey guys, just checking in so you know I didn't forget about this! @pmvcda you picked an excellent time to do your rewrite, as I'm too busy to test much right now. I still have a lot of work to complete to finish up this cockpit, it will be a couple weeks at the least. Once it's done though I'm going to be back to trying to get smooth heave and pitch movements. Most everything else is good with Mover 1.3 already.

I do still think that the software filters interact strangely with the AMC1280 though, I don't have enough knowledge to understand but maybe Dirty can figure it out when he gets back from holiday.

Oh, I can figure stuff out especially well while at the pool

As far as I could tell, the filters in FlyPT_Mover work great. At least the EMA highpass, lowpass and their combinations. What interaction can there be? The software tells the AMC "where to be" and the AMC takes care of "how to get there".

Can you try to describe what exactly you observe?

Dirty

That makes sense to me, if I'm stuck spending 'recreational time' instead of working on my hobbies I do my best to at least work on my hobbies in my head. =D

When using low pass filters it makes slow movements really jerky. Each actuator starting and stopping moving at random intervals instead of moving slow and steady to the setpoint.

With somewhat faster (but still slow) movements it's just shaky, like low frequency vibration or small rapid oscillation maybe. I can try to get a video but it will be a couple days before I have time, we have family in from out of town again.

It's bad enough to prevent me using the LP filters strongly enough to smooth out my sim (mostly those pesky pitch and heave movements) really effectively. I'd have to look back through my posts to find it but I posted previously about a more exaggerated symptom of this.. I think it was that if I set any significant amount of "I" (maybe in the ballpark of 10 I think) in my PID on the AMC settings, with significant software filters running on the PC at the same time the platform is very very unhappy with some fairly violent oscillation problems. Or was it "D"? I don't know something along those lines that confirmed to me that the jerky movement and shaking really were an interaction of some kind between the two because they went away as soon as I turned off the software filters or set "I" (or maybe it was "D") back to zero. Don't trust me 100% on this stuff, it's been probably over a month and I don't trust my memory at all. You really have to "feel" the issues I have, to a great extent they are too small to see but drive me nuts!

Also when @pmvcda gets that source created to allow me to crop the incoming signals (like you can do with SimTools tuning center) that will probably be a huge help. For the Huey I think I'm going to run your customized DCS plugin instead of FlyPT Mover, but I absolutely love Mover's features (especially the super customizable on-ground/in air solution) and the layout. With the Huey the "on ground" customization is less important, though I do really like with helicopters to also use bigger roll and pitch when "on ground" to better feel what exactly is happening during "pick up" and to increase the touchdown bump. I would need you to make a way for me to use the helicopter cockpit G forces values if you haven't already, I don't recall for sure but I think maybe you already did...

Also @Dirty thank you for the reply about my heave being not straight up and down. What you said is the same thing I was thinking, that it means somewhere my measured dimensions are off. Mostly I just wanted to make sure that this isn't normal for a Stewart rig before I go crazy trying to get more accurate measurements of everything to find any errors. I know the top platform is very accurate but the bottom ones are likely off a little bit. I'll check the rods too.

Thanks!

From what you're saying, I'd look into the PID tuning. As far as I can tell, the processing and filtering of FlyPT_Mover are fine. I don't think that the actual signal is shaky, but most probably the conversion of the signal into a position in the actuator.

Dirty

Hi everyone,

Sorry for being away... but work is progressing. Not as fast as I wanted.
Some new features coming with the next version.
Updated graphics and 3d models, new rigs, new sources and advanced filters.
Some big internal changes on actuators, to allow user defined rigs in the future.
Code seems to run fast with calculation and updates under 2 ms. Seems good for now, at least on the PC's I tested.
Trying to reach the end. Really curious to get feedback on the updates.

Looking good

Looking forward to this @pmvcda
For the loop function, could you make a feature that allows a different cycle frequency per axis chosen. So we can get the rig rocking around like a boat if we picked say 1 second for the roll and 1.8 seconds for the pitch?
Currently if I select pitch and roll it see saws from forward right to rear left.

You can already, just add more sources.
Set each source to the loop you want and select were you want to use each source.

New advanced filtering system:



Now we can compose the filter with multiple operations and specific values.
This allows users to crop/limit values received like @Trip Rodriguez requested.
And also play with a multitude of filtering solutions.
Filter can be generated in the combo boxes visually or by writing the code in the text box.
Here's a quick reference of the filtering system (might change till release):



ADVANCED FILTERING

Advanced filter is composed by a series of operations.
Each operation can have parameters.
Those parameters can be numbers or other operations.

Possible operations:


VALUE
Is the value received.
No parameters available.


NUMBER
To specify a numeric value.


EMALP
Exponential moving average low pass filter.
Parameter 1 is the value we want to filter.
Parameter 2 is the number of samples used in the filter.
Use this to smooth a signal by removing high frequency variations.
Only low frequencies pass (look at figure bellow).



The number of samples define where we have the cutoff.
Cutoff is not a vertical line, only in ideal filters this happens. Reality is a progressive change like the curve shown in the figure.


EMAHP
Exponential moving average high pass filter.
Parameter 1 is the value we want to filter.
Parameter 2 is the number of samples used in the filter.
The high pass filter is the value we want to filter less the low pass filter for the specified samples.
So the number of samples specified, define the low pass filter used.
Here, only high frequencie values pass.
Low frequencie changes/smooth changes are removed, depending on the selected number of samples (see image bellow).



The formula for this filter is:
EMAHP = VALUE - EMALP(VALUE)


EMABS
Exponential moving average band stop filter.
Parameter 1 is the value we want to filter.
Parameter 2 is the number of samples used to define the start of the band (filter #1).
Parameter 3 is the number of samples used to define the end of the band (filter #2).
This filter, removes all the frequencies between filter #1 and filter #2.
So we get low frequencies and high frequencies.
Filter #1 and filter #2 define the start and end of the band of frequencies we want to remove.



The formula for this filter is:
EMABS = EMALP(VALUE) + EMAHP(VALUE)


EMABP
Exponential moving average band pass filter.
Parameter 1 is the value we want to filter.
Parameter 2 is the number of samples used to define the start of the band (filter #1).
Parameter 3 is the number of samples used to define the end of the band (filter #2).
This filter, only allows frequencies between filter #1 and filter #2.
So we remove low and high frequencies.



The formula for this filter is:
EMABP = VALUE - EMABS(VALUE)


DEMA...
Double exponential moving average filters.
Like EMA filters, but "2nd order".
Compared to EMA, those filters have less lag, but they also overshoot (see TEMA figure bellow).
Formulas are:
DEMALP = 2 * EMALP(VALUE) - EMALP(EMALP(VALUE))
DEMAHP= VALUE - DEMALP(VALUE)
DEMABS = DEMALP(VALUE) + DEMAHP(VALUE)
DEMA BP = VALUE - DEMABS(VALUE)


TEMA...
Triple exponential moving average filters.
Like EMA filters, but "3rd order".
Faster response compared to DEMA, but even more overshoot (see figure bellow).
Formulas are:
TEMALP = 3 * (EMALP(VALUE) - DEMALP(VALUE)) + EMALP(DEMALP(VALUE))
TEMAHP= VALUE - DEMALP(VALUE)
TEMABS = DEMALP(VALUE) + DEMAHP(VALUE)
TEMA BP = VALUE - DEMABS(VALUE)


LOGISTIC
Logistic is a case of the sigmoid filter.
Here we use a faster simplification.
Parameter 1 is the value we want to filter.
Parameter 2 is the range (minimum/maximum value allowed).
Parameter 3 defines the S shape. High value makes is a more vertical slope while low values make it more horizontal.
Use this filter to fit values to a specific range
Reaching and surpassing the maximum range, values are "compressed" to fit the specified range.
The S slope defines how fast and how much we compress the values.



CROP
To define minimum and maximum for a value.
Parameter 1 is the value we want to limit.
Parameter 2 is the minimum value we allow.
Parameter 3 is the maximum value we allow.


REMAP
To remap a value from a range to another range.
Parameter 1 is the value we want to remap.
Parameter 2 is the start of the range where we have the received value.
Parameter 3 is the end of the range where we have the received value.
Parameter 4 is the start of the range where we want to have the received value.
Parameter 5 is the end of the range where we want to have the received value.


GAIN
Is a multiplication operation.
Parameter 1 is the value we want to modify.
Parameter 2 is the gain we want to apply.
Use this to invert values by using negative numbers.


TRANSLATE
Is an adition operation.
Parameter 1 is the value we want to modify.
Parameter 2 is the amount we want to move the value.


ABS
Get the absolute value.
Parameter 1 is the value we want to turn into absolute value.

@pmvcda AMAZING!!! I can’t wait to get stuck into this aspect of the new version! Thank you.

Working on the pose calculation...
So many little things causing bugs!
Seems to be good now:



2ms calculation loops (lower right corner of main gray window)
Showcasing the new filter control.
And a small source change. No more gravity option. It's now on the list of values available. So we can choose acceleration with or without gravity included.

It's getting closer to the new release.
Big internal changes and reorganisation because of threads.

Almost there!

And some new rigs:

Sorry for interfering but for the costistency and user’s understanding I wouldn’t make a parallel between:

DEMA and second order EMA and
TEMA and third order EMA

as those are totally different things in my opinion.

2*EMA is not the same as EMA of EMA