Fitbit app and Sense 2 show different distance

[Jeetthaur]

Hello everyone,

I walked a distance which is measured by different ways like another smart watches , bands & car etc. but there is huge difference between sense 2 measurement & other device’s measurement.

Even distance measured by Fitbit app & same distance measured by watch is different.

f.e.- a walking session i start in watch & same session starts in app..

when i complete 1 km distance in app that  time watch shows 850 mtrs.

when i end this session… fitbit app shows 3 kms. & watch shows 2.55 kms.

why..?

Moderator Edit: Clarified subject

[LZeeW]

Hello @Jeetthaur and welcome to the Community.  There are hardware and software differences between your Sense 2 and using the Fitbit app on your phone.  This is true if you compare it to any other device.

• The GPS antenna in your Sense 2 is very different from the one in your phone.  You use the GPS antenna in your phone when you track your distance in the app.  Different devices have different accuracy.
• The sampling frequency is probably different between any two devices.  A device that samples more frequently is generally more accurate.
• A car GPS might have programming that will put you on the road, even if you are off the road.  You can't compare your Fitbit or your phone to a car.

[Guy_]

Hi @Jeetthaur - the probable reason for the difference is the phone GPS is more robust and the tiny GPS in the watch has been known to lose signal and might not compensate for the lost portions.

You can possibly check this by walking the same route on several occasions and if you get different distances, that is a likely explanation. Assuming you don't get identical dropouts due to buildings or trees.

[t.parker]

It doesn't come down to the GPS. For step-based activities, Fitbit uses a different algorithm to determine the distance. This can be seen in the TCX files where the distance doesn't follow the GPX coordinates. It makes sense as walking/running is quite a slow (compared to vehicles and bikes) activity and the accuracy of the GPS isn't enough to provide real-time pace and distance and correct GPS errors at the same time. Most manufacturers (maybe even all) use inertial positioning as well as GPS (if available). So that includes using a majority of sensors to compute the metrics. Each sensor has its own issues (for example, the accelerometer suffers from drift, an integration error which very quickly adds up to the distance and must be corrected). The more sensors used the better chance that one sensor will serve to correct another one and so on. This is well visible on Garmin watches when using in-watch navigation and despite losing the GPS signal for a brief moment user is still being navigated on the map using inertial navigation (which eventually will break if GPS doesn't restore a connection in time). Fitbit's sensor array isn't very wide (and with new devices, some sensors are gone like gyro, also there is no magnetometer which may indicate north etc.) so the overall accuracy may be lower. It shouldn't be however that low even when using GPS with a lower sampling frequency (and the sampling frequency is 1s, see the TCX file, nothing different from other devices). The GPS waypoints eventually should serve to correct distance issues. The algorithm should be able to tell whether the waypoint is an error. For example, the user isn't moving but GPS is jumping all over the place (this is visible in TCX data when the activity is paused), or when GPS starts drifting, so the watch should use steps, possible strides and say "Hey, that's too far, too fast, not possible for a human on foot, GPS must be wrong!". This is how most watches work that out and discard potential errors. Using GPS only is a bad idea for step-based activities when we want to see real-time metrics updated every second.

One thing I notice is that Fitbit doesn't seem to work anymore on the distance calculation algorithm. Since Charge 2, I see no improvement in both, the accuracy and consistency of distance calculation. Lack of consistency is more worrying. It is fine if my Parkrun ends with 4.80km as long as it does that every time (let's say 4.80, 4.81 - small differences would be normal). Instead, I have the same route captured with distances 4.80km, 5.06km, and 2.11km (no GPS dropout on this one, not sure what happened). I have a theory that if we use automatic stride length, the GPS contributes to it and previous run/walk with unstable GPS may trash that value for the next activities but I haven't 100% confirmed it (I have some evidence that indicates that but I haven't been tracking it continuously so it could be just a coincidence).

Fitbit's distance computation has been a problem for a very long time (there are posts years old here) and nobody knows how it's calculated. I can tell how it should be done (from my own work with navigation and from other manufacturers' papers). Walking/running distance is quite complex to get right and I don't think any brand has it 100% always working. Even when using mobile phones, I know that Android has quite a complex algorithm in place for determining positioning and it doesn't rely on the GPS only. However, I also have a feeling that since Fitbit developed its algorithm many years ago, it stopped improving it and the same thing is being transplanted into the new devices.

Pixel Watch users may be more in luck as they may use foot pod sensors and other software to get distance (and some may integrate with Fitbit).