[chbot] Filter

[Mike Field]

I got a copy of the data too... If you want to play around with it, here 
is a Google Doc Sheet, that is set up for a FIR filters with 19 taps

https://docs.google.com/spreadsheets/d/1PR5CtZAdveY66vysHhGz3SXujTjBPddhRq1XDc27KHs/edit?usp=sharing

To me, it seems like slightly modified a 12-point moving average seems 
to make the best velocity curve (where best = smooth). A simple moving 
average was modified to make it symmetric, so it doesn't shift the 
values half a point to the right.

But in general, because the noise is not independent of the signal, and 
it doesn't look to be symmetrical, it it very likely to have low 
frequency components that can't be completely filtered out.

Mike


On 20/10/2018 5:17 PM, Eliot Blennerhassett wrote:
> On 20/10/18 1:46 PM, Trevor Wignall wrote:
>> A quick look at the individual data points suggests that there is a
>> regular higher reading every fourth reading.
> These jumps appear to be proportional to the slope of the graph. They
> are higher on the ascending side, and lower on the descending side.
> You can see it here
> https://github.com/eliotb/jupyter-notebooks/blob/master/cam_analysis.ipynb
> in the second graph.
>
> It could be from the rotation being uneven every 4th step...
>
> Pretty well removed with low-pass filtering.
>
> LPF seems to be the best; median and averaging filters don't work so well.
>
>
>
>
>> Without knowing more about
>> the sensor system, I cannot guess whether this is an artifact of the
>> sensors, or some form of interference. However if this is confirmed, it
>> may be possible to measure this stray signal - whatever its cause - and
>> subtract it from the data. I am assuming that this isn't the actual data
>> that you are seeking to measure of course!
>> It would help to know what information you are trying to extract from
>> this data, and what can be ignored, so we can suggest appropriate
>> filtering techniques that leave what you want and remove what you don't
>> want, rather than the other way around :)
>> Trevor
>>
>> On Fri, Oct 19, 2018 at 2:55 PM Eliot Blennerhassett <ewblen at gmail.com
>> <mailto:ewblen at gmail.com>> wrote:
>>
>>      On 18/10/18 3:27 PM, Eliot Blennerhassett wrote:
>>      > Can you share the actual raw data?
>>      > Preferably from say 10 continuous rotations.
>>      >
>>      > Real data trumps hand-waving hypotheticals any day.
>>      >
>>      > Are the measurements at e.g. 1 degree intervals, or at equal time
>>      > interval? etc?
>>
>>      Wallace sent me a data file, here's a first look at it, and low pass
>>      filtering of same:
>>
>>      https://github.com/eliotb/jupyter-notebooks/blob/master/cam_analysis.ipynb
>>
>>      _______________________________________________
>>      Chchrobotics mailing list Chchrobotics at lists.ourshack.com
>>      <mailto:Chchrobotics at lists.ourshack.com>
>>      https://lists.ourshack.com/mailman/listinfo/chchrobotics
>>      Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
>>      Meetings usually 3rd Monday each month. See http://kiwibots.org for
>>      venue, directions and dates.
>>      When replying, please edit your Subject line to reflect new subjects.
>>
>>
>> _______________________________________________
>> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
>> https://lists.ourshack.com/mailman/listinfo/chchrobotics
>> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
>> Meetings usually 3rd Monday each month. See http://kiwibots.org for venue, directions and dates.
>> When replying, please edit your Subject line to reflect new subjects.
>>
>
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
> https://lists.ourshack.com/mailman/listinfo/chchrobotics
> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
> Meetings usually 3rd Monday each month. See http://kiwibots.org for venue, directions and dates.
> When replying, please edit your Subject line to reflect new subjects.