Pull in housekeeping + playback mode on PC side

[tygamvrelis]

Minor changes:

• Fix units typo in tables printed out during recording
• Interpolate nans when loading log files for analysis and playback
• Changed --log script argument to --record
• Fix script argument parsing bug for booleans
• Fixes for Raspberry Pi:
  • Gracefully handle failure to import matplotlib
  • Use "lamp_data.dat" as base filename instead of date and time IF Linux is detected
  • Use "/boot/lamp_data" as data directory IF Linux is detected

Major changes:

• Add ability to use a run for calibration. The calibration settings are stored in settings.ini so that they can be quickly retrieved during playback and visualization
  • This works as follows: first, compute the median value for each of the accelerations. Next, multiply by -1 to flip axes as necessary. Finally, compute the error vector by comparing against the steady-state measurement vector [-9.81, 0, 0] (z, y, x). The deviations from this vector are the additive calibration offsets, and the -1 factors are determined by which way the local x-, y-, and z-axis point for each IMU
• Add playback mode to Python script (may not completely work!)
  • Need to implement embedded side of this, but the PC side is more or less there

Next steps:

• It turns out that all the IMU acceleration values are off by a factor of -1. This means that there needs to be a change in the embedded code that multiplies ax, ay, az by -1. This will make the acceleration values received by the computer log consistent with the coordinate system described in the sensor datasheet. Once this change is made in embedded, the multiplication setting in settings.ini needs to be changed to: mult_z = -1, mult_y = -1, mult_x = 1. Which is the negative of what it is right now. This will not solve any functional issues per se (currently, two wrongs are making a right!), but it's a good improvement for consistency
• Implement playback on embedded
• Do a few hands-on recording tests to make sure the angle conventions are as expected

[tygamvrelis]

I want to do a recording test before pulling this in just to make sure my changes didn't break the recording module. I can do this at home; it just require a MCU, not the lamp setup. Will try to do this on Monday or Tuesday.

[bturep]

Sounds good, ty Tyler!

On Sun, Jun 30, 2019 at 2:36 PM Tyler Gamvrelis notifications@github.com<mailto:notifications@github.com> wrote:

I want to do a recording test before pulling this in just to make sure my changes didn't break the recording module. I can do this at home; it just require a MCU, not the lamp setup. Will try to do this on Monday or Tuesday.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/tygamvrelis/oil_lamp/pull/22?email_source=notifications&email_token=AMDHDXFYNLLNTY4YSLXVKV3P5D4K5A5CNFSM4H4LTMW2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODY4RKOQ#issuecomment-507057466, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AMDHDXHAWXXUAHHOMA6ET2TP5D4K5ANCNFSM4H4LTMWQ.

[tygamvrelis]

Ran a few recording tests using the same MCU code as in master, looks good.