Gyroscope and accelerometer measurements of the human body.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data_collection
data_processed
figures
led_visualization
.gitignore
Makefile
README.md
align_data.py
f_transform.py
plot_time.py

README.md

Trapezium

Gyroscope and accelerometer measurements of the human body.

Data collection

Sample data output:

3 452299760 15144 -7869 -606 -25 31 22
0 452244271 15099 -7581 -4928 9 26 41
1 452188739 -12989 2879 9970 199 -202 -69
2 452133211 16079 2566 2249 -91 33 68
3 452014296 15089 -7938 -585 -21 30 21
0 451958762 15108 -7594 -4867 11 29 42
1 451903483 -13102 2701 10252 201 -180 -71
2 451763466 16127 2764 2145 -46 20 87
3 451510744 15248 -7903 -496 -25 28 20
0 451455242 15128 -7554 -4902 10 27 46

Field order

device timestamp ax ay az gx gy gz

Notes on the data fields:

  • The timestamp is the output of a "binary counter: attached to the CPU clock. In fact, it is a 32 bit binary counter, and the CPU clock speed is 200 MHz. It counts down from 0xFFFFFFFF and when it hits zero it repeats. A difference of one timestamp = 1 / 200*10^6 seconds.
  • Each sensor should be outputting data at 400 samples/second.
  • If you want to convert the gyrometer data to "degrees/sec", multiply by 2000 and divide by 2^15.
  • The accelerometers always have a "DC bias" induced by the acceleration of gravity. To convert from the integer-valued acceleration to units of "gs" divide by 2^14.
  • The accelerometers "clip" at 4g's or a maximum integer value of +/- 2^15. For fast activities such as sprinting we have the option of changing the ceiling to 8g's or even 16g's.
  • Notes on power lights: 1] is streaming computer 2] is recording to flash memory 3] is data already written to flash memory (that is, do we need to clear it)

6/25/17 First field test

Took several measurements walking, jogging, running, sprinting, backwards walking, jumping, squats, and stairs. Sensor device order (0-3-1-2) mounted roughly on the waist, upper thigh, top of knee, and above the ankle.

Sample measurements should have taken approximately these time intervals:

  • setup ~80s
  • walking 53s
  • jogging 26s
  • sprint < 8s
  • setup 15.5s
  • jumping 23.5s
  • squats 15s
  • setup 33s
  • downstairs 15s
  • upstairs 20s

Segmented walking data

Waist

Sample Data

Upper Thigh

Sample Data

Knee

Sample Data

Ankle

Sample Data

Segmented walking data modeled with a Guaussian Process

Waist

Sample Data

Upper Thigh

Sample Data

Knee

Sample Data

Ankle

Sample Data

Full dataset plotted for lower leg

Sample Data

Visual shots

https://twitter.com/metasemantic/status/881317564680478720