Add elliptical machine profile

[ghost]

Hello, I am very fresh user (installed today my first OpenRowingMonitor) but I also have an older elliptical machine that my SO is using. I was wondering, if it would be possible to implement such profile for elliptical machine (a.k.a. orbitrek). I guess it is similar to indoor bike.

[JaapvanEkris]

That is a tough one. One first sight, it would depend much on how the elliptical machine's flywheel behaves. ORM needs a recovery phase, as it uses this to determine the drag factor. This could be fixed (autoAdjustDragFactor set to false), but many metrics are only calculated when a stroke is detected somehow. Do you have any idea how the flywheel behaves?

[ghost]

I guess the flywheel is mechanically connected with the platforms. So no spindown is occuring. The only way to have the drag factor data, would be to intercept the communication from original computer to the motor that adjusts the magnetic resistance, or use some kind of linear volume slider attached to the moving arm of the regulator?

[JaapvanEkris]

Well, a rower calculates the drag during the recovery, when there is no power on the flywheel. A concept2 BikeErg has a function to recalibrate the drag: you need to spin up the flywheel, and then let it spin down.

The issue is that most of our calculations (which can be found in RowingEngine.js) are based on the rowing stroke: some metrics aren't even defined for rowing without having the concept of sroke. For example: power is defined as the cube of the "average angular velocity over a stroke" multiplied by a constant. This is what Lars and I call a robust calculation: it isn't easily affected by a single bad measurement. There is another calculation that is insensitive to the concept of a rowing stroke, where you have to take in account the difference in angular acceleration across measurements. This is a calculation that isn't robust and thus is vulnerable to noise in the data.

Such alternative power calculation is available in the newer rowingengine I am currently finalising (as InstantPower, as a preperation for games Lars is designing), but it hasn't been tested on other machines than a Concept2 RowErg. Please note that this involves some pretty hefty maths and physics, where I am using Langrange theorem combined with full Theil-Senn regression to obtain a robust estimate of angular velocity and angular acceleration. Even then, we keep using the robust calculation for power on the GUI and we only use these metrics to calculate the PowerCurve and ForceCurve of a stroke. And I already know that due to some mathematical properties the InstantPower will not line up with robust alternative (it is close, but not enough for displaying to a user). Full Theil-Senn and Lagrange require significant CPU power, where I am far from certain that it will deliver timely results on lower powered devices like a Raspberry Zero. Alternative less power consuming algorithms (a partial Theil-Senn that isn't as robust to noise) work OK on my Concept2 but they might work better or worse on other machines depending on the noise level. So this isn't a trivial task.

From there on, getting from generated power on a flywheel to linear distance/speed isn't trivial as you need a magic constant: how much distance will a rotation provide? The formula's might resemble the rowing machine (see http://eodg.atm.ox.ac.uk/user/dudhia/rowing/physics/ergometer.html), but it would require some investigation about how energy is transformed in a linear fashion.

[laberning]

I agree. An elliptical machine works similar to an indoor bike.

During the workout the wheel spins with an almost constant speed and the force required to keep the wheel spinning is controlled by some kind of brake (i.e. mechanical or magnetic damper).

That's actually quite different from what a rowing machine does. I'm note sure about your exact use case, but FTMS has a special profile for elliptical machines that might be used if you want to connect it to workout apps via BLE.

Maybe there is already open source software out there to do something like that. Otherwise what you probably would have to do is measure the damper setting of your device and then measure the speed of the wheel. This would give you the power (quite similar to indoor bikes).