01-15-2021 09:44
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post

01-15-2021 09:44
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post
Hi - Novice here. I am interested in logic that would distinguish between the states of "driving (in a car for example)" and "not driving". I understand that I can sample and smooth accelerometer data, but I am really struggling with it. Has anyone done this before? Can someone point me towards a resource? TIA.
01-18-2021 05:40
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post


01-18-2021 05:40
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post
Fitbit don't publish their algorithm, but there are lots of articles which discuss how to use an accelerometer sensor to do step detection. Once you understand how steps are detected, you should be able to see how driving vibrations might differ from those.
An article like this might help: https://www.analog.com/en/analog-dialogue/articles/pedometer-design-3-axis-digital-acceler.html

01-27-2021 05:25
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post

01-27-2021 05:25
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post
If targeting a device that supports location API, I would use the GPS to calculate speed to determine that a speed > xx = driving.

06-28-2021 07:13
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post

06-28-2021 07:13
- Mark as New
- Bookmark
- Subscribe
- Permalink
- Report this post
Can you elaborate? Thanks for replying.

