Accelerometer Deep-Dive

Summary

Let’s take a more in-depth look at the accelerometer. As a reminder, when we talk about accelerometer magnitude, this means the vector magnitude of the force on the accelerometer in 3D space, as given by this formula.

Take a look at this accelerometer trace when the wearer is at rest.

The accelerometer magnitude is close to 1 g, which means the total force on the accelerometer is only due to gravity. But if you look at the individual channels, they shift around. In the beginning, they all see the same amount of gravitational force, which means the device is tilted along each axis equally. But then after 15 seconds, the x and y channels drop and the z-channel rises to nearly 1g. This indicates the device has moved and is now lying flat, so only the z-channel feels the force of gravity. By strategically placing accelerometers on the body, you can use this technique to figure out things like posture or sleeping position.

This is a trace of someone who is jogging. Notice the very periodic shapes in each channel that indicate the cadence of the arm swing, which can be used to figure out how fast someone is walking. By analyzing the specific characteristics of the waveform, you can decompose each stride into various components.

For example, this paper decomposes the accelerometer trace from an ankle into various phases of the gait cycle.

Discriminating the gait phases like this is an important starting point for several applications such as recovery after an injury or treatment, the classification of daily activities, coaching athletes, and distinguishing between normal and pathological gait. Researchers have found that differences in each person’s gait mean that it can be used to identify individuals.

Here we see a zoomed-out view of the accelerometer trace over 2 minutes. We can see the wearer distinctly changing activities, going from jogging to still to jogging again. You can even tell that they are probably jogging on a treadmill as the speed slowly ramps up and down. And from the sharp cut-off around 50 seconds, maybe we can tell that they jumped off the treadmill or for some reason their treadmill session was cut short. I didn’t collect this dataset, so we can’t know for sure, but it’s interesting that we can learn this level of detail from just an accelerometer alone.

Quiz

Review the set of graphs to answer the following quiz.

QUIZ QUESTION::

Match the accelerometer traces above with the activity being done in each of the above image.

ANSWER CHOICES:

Exercise 1: Step Cadence

Instructions

1. Complete the Offline or Online instructions below.
2. Read through the whole .ipynb.
3. Complete all the code cells that contain ## Your Code Goes Here.

Offline

1. In the repo which you can access here in the repo /intro-to-sensors/exercises/1-step-cadence/ you should find the following files:

• 1_step_cadence.ipynb
• exercise1.npz

1. Open up the python notebook and associated files in your desired editor.

Note: Instructions can be found in Introduction to Wearable Data's Concept Developer Workflow for how to set up your local environment.

Online

1. Go to the next concept and the 1_plotting.ipynb should be open and the workspace should already contain the appropriate exercise1.npz file.