Why running wearables fail: behind the scenes of our smartwatch demo

Last week we launched our smartwatch demo app that demonstrates how our technology can enable extraordinary gains in accuracy when embedded into standard wearables - to prove it we ran two demos on a track and on a road course.

We wanted to peel back the technology and tell the story of what’s going on inside those tests - and why regular wearables can be so inaccurate in recording position and path.

Our Senior Software Engineer, Henry, track- and road tested the technology in two ways:

Test #1: Accurate distance travelled

For this, we simply ran a standard 400m track at Battersea Athletics Track. This test shows the side-by-side results from a Fossil Gen 5 watch - the GNSS data that the watch uses, against the result that it can achieve using our D-Tail technology.

Sports watch test on a track

On a track which is fairly well exposed to the open sky you would expect that you might get a good GNSS signal however, the surrounding trees and a small metal-clad grandstand may have obscured some satellites signals or created reflected signals.

In addition, despite saying that it had a fix, the watch may not have acquired enough satellites to provide accurate positioning from GNSS alone. In cases like this, FocalPoint's Human Motion Model enables us provide a more accurate trace of the body's motion through space.

Finally, the GNSS antenna inside the watch is usually smaller than you would expect to have in a smartphone or in-car satnav, making it less sensitive, and sometimes the dynamics of running itself, and the shocks and accelerations on the device can impact the signal to noise ratios of the received signals.

All told, it's no surprise that using standard GNSS alone, the watch only managed a distance reading of 473m for a lap of a 400m track. By comparison, using FocalPoint's technology, it measured 396m

Test #2: Accurate line on a map

For this, we ran through the streets of Clerkenwell in London to demonstrate how our Human Motion Model can accurately capture movement to a cm-level, even in GPS-denied spaces; and how we can filter out bad GPS fixes to increase the accuracy of the path.

Route paths from street run with and without FocalPoint

When you zoom into the route, you can quite clearly see the reasons why current sports wearables struggle to provide accurate view of the route that you have run:

Error: Signal bounce As we run along a narrow pathway, signals from the south are bouncing before being received, causing the device to assume it’s further from the truth. Error: Sudden change in the constellation As we run along an east-west road the GPS receiver loses the north-south satellites in the sky. When we emerge onto a north-south road the error is reversed and the sudden change causes a major error in positioning. Error: Signal bounce As we run along a city street, satellite signals bounce off buildings, causing incorrect positioning fixes (red). FocalPoint’s human motion modelling software (blue) checks positioning against human motion and rejects incorrect fixes until confidence in GPS is restored. Error: Bad first fix Over the first few seconds, the wearable (red) picks up more and more satellites, causing the positioning fix to jump around as it changes its estimate. FocalPoint algorithms (blue) are tailored to discard incorrect satellite fixes and improve the stability of the first fix.

A deep dive into why GPS errors occur, using the lines from our smartwatch demo city run

It's clear from the infographic above that the architecture and environment of modern life makes it harder for GNSS receivers to achieve an accurate fix, creating errors from the wearables and smartphones that they're inside of.

It's not just cities and urban canyons either: Geographic features and even trees can obscure or reflect GNSS signals too - meaning that a run in the countryside in the winter versus in the summer could return significantly different distance and route readings too.

We’d love to hear your feedback on the demo. Perhaps you have some other wearable fails which you can share with us? Please tag us in your social media posts so we share them.

Fellow runners - help share our demo film on Instagram and Twitter.

Manufactures/Developers - request a demo here.

To learn more about the above use case, please head to our Running page.

Useful links:

  • Why you can’t trust Strava to help you run your backyard COVID marathon [read here]
  • Here’s how badly your fitness tracker is lying to you [read here]
  • Listen to one of the Run Testers, Kieren Alger, discuss our technology on the Runners World UK podcast [head to 17mins in]