Last Updated on December 11, 2022 by pg@petergamma.org
If we look at the data from Rob ter Horst, who measures step counting accuracy, we cannot find any clear answer.
We found, that a combination of a Garmin Venu watch, a Suunto foot pod, the Android app a Training tracker and a manual step counter can deliver us a lot of information about how we walk.
We do not know of any step counter which is 100 percent accurate. We did some tests with two Venu watches, two foot pods and two Android phones running a training tracker. When a foot pod is paired to a Venu watch or to a training tracker, then we find, that both devices offer the option to calibrate the foot pod. But the calibration does not affect step counting, but only the other parameters.
When we compared the data from two devices, we found, that there is a scattering in the data. A Garmin Venu watch can deliver us a lot of information about our walking activity with the protocol indoor walking. The protocol counts the number of steps during in an activity. Unfortunately, the number of steps during an activity can only be seen on the watch face, but does not show up in Garmin Connect.
Once upon a time, the protocol indoor walking from Garmin showed the total steps during an activity in Garmin connect, as well as lap steps. But these data are not there anymore in Garmin Connect? Is this because users complained about inaccurate step counting?
We can use in addition to a Garmin watch with a foot pod the Android app a Training tracker with a foot pod. The Garmin Venu watches warns us, that if we walk at a speed which is slower than 90 or 100 steps per minute, that the step rate is too low. If we ignore this warning, we can see that in Garmin connnect, that at slower speeds, the step rate drops to zero.
A training tracker records the number of steps to a .csv file. But if we walk at speeds between 90 to 100 speeds per minute, we can see that there are drop outs in the data of a training tracker. We observed this in two devices independently.
Garmin does not allow any analytic with steps in Garmin Connect. We can record the steps with a training tracker. We can use additionally an Apple watch, which counts steps quiet accurate. These steps can be synchronized from an Apple Watch to Home Assistant and InfluxDB, as described in this blog. In InfluxDB, we can do some analytics with Python scripts.
To increase the step counting accuracy, we could use several devices which are wrist based with foot pods. We could not find a step counter which is especially accurate. But a combination of several devices could increase the accuracy.
If we have the option to do some analytics with our step data, as this is offered by Apple watches, who have the option to synchronize step data into Home Assistant and InfluxDB, and do some analytics with it, we are convinced that it is possible to get very accurate step counts. The example of the protocol indoor walking demonstates, that we can get a lot of information about how we walk with a sports watch.
Unfortunately, Garmin strongly limits what users can do with their data. Newer devices like Apple watches, Home Assistant and Influx DB have more options to do some analytics with our sports sensor data, as described on this site. It is highly desirable, that these options are developed further, so that users can do their own special analysis with Python scripts. Every user should be free to do with their sports sensor data whatever they want to. Eventually, skilled data scientists can increase the step counting accuracy, if they have the necessary tools to analyze step counting data from several devices.
According to our own little experience, more parameters than only step counting should be used to analyse walking meditation, like speed, distance, step rate and stride length. These data can be optained from certain sports watches. Rob ter Horst is satisfied with the Apple watch 6 for step counting. But we do not know of any tools to get these parameters from an Apple watch 6. The Android App a Training tracker from Dr. Ing. Rainer Blind, an open source app which can be found on Rainers Github page, offers to pair many sensors simultaneously. We can pair for instance four foot pods on our feeds and on our hands. In that way, we get a lot of information for scientific studies. A Training tracker also offers to snyc data to Strava. These streaming options should be developed further.