I’ve begun to experiment with the speed attribute of the device tracker associated with my mobile phone. This is a journey I did via car (the entity is a template sensor that wraps the speed attribute):
Although the “shape” of the chart makes sense, I am almost certain I drove faster than the peak of 5m/s (or 18km/h or 11.2 mph). This “factor” doesn’t seem to scale either, with highway driving (110km/k or 70mph) peaking at around 8m/s (according to the value of the attribute).
Is this a known issue with mobile phones being used as device trackers? It’s a Pixel 6a if that means anything.
The time between location updates is very low. So the average ofer that period includes both times you do not move. It also assumes movement in a straight line between two updates. Driving around in circles could result in speed 0.
That makes sense in the general sense, although these were pretty much “straight line” trips with updates that appear to have been made every 5 seconds or so, so I wouldn’t expect this much of an error unless there’s something wrong.
Also I presume that apps on the phone would suffer the same thing, but they don’t (even basic GPS apps have the correct speed).
I think I had the incorrect units specified in the helper pulling this attribute out. I didn’t realise it would manipulate the result but recreating it with the correct units appears to have fixed it to more sane values.
