/ suunto

Thoughts on inertial navigation, and a week with the Suunto 9

Inertial navigation is a wonderful thing: through an elegant system of gyroscopes, accelerometers, and magnetometers one is able to measure movement in space without an external source of reference.

https://en.wikipedia.org/wiki/Inertial_navigation_system

The technology is mature, and used regularly: even the latest commercial aircraft have several discrete INSs to backup their GNSS in the event of failure. If you look at the aircraft stand at the airport, you can see coordinates that the pilots can use to calibrate their INS prior to take-off.

Keeping with the aircraft use case, the designer has two pieces in their favour: available space, and predictability of motion. Under normal circumstances, an aircraft - especially a commercial airliner - moves smoothly, allowing for an easier calculation of course. Where it hasn't been significantly used is in the mobile or wearable space, although this iOS app looks interesting:

With the 9, Suunto has taken their (now-polished but much-maligned) Spartan platform, added a new GNSS chipset from Sony, and much to the delight of the self-quantification community added a new method for tracking one's progress through space.

Long observers of Suunto will not be surprised: they introduced both FusedAlti (allowing GPS and a barometric altimeter to work together to keep calibration true) and FusedSpeed (allowing speed to be determined whilst running from a calibrated stride distance and cadence detected by the watch), key building-blocks for FusedTrack. Short-form, the watch updates its known location from available GNSS once a minute, saving valuable battery power during the interval. In the gap between these minutes, the 9 uses its plethora of sensors to determine speed, heading, and distance. This in turn allows for tremendous battery life, and the scope to have more accurate walking and outdoor swimming tracks as the technology matures. I get the feeling that as of firmware 2.1.54, the 9 isn't quite as accurate as the WHRB I have running 2.0.42, but it's well within an acceptable margin (e.g.: it records where I crossed the road with slightly less elegance than the WHRB).

GNSS track of a Suunto 9 crossing a road overlaid on a satellite image
The 9 tracks the road crossing with an agreeable degree of accuracy: unforutnately for the watch, both sides of the road are under power and telephone lines, and I wear it on my left - building-side - wrist.

In 2012, Nokia were announcing their indoor navigation, with a claimed accuracy of 30cm, but this required a network of external beacons. A subject that has not been discussed is the monumental step forward represented by accurate inertial navigation, in such a small platform. Like any sensible company, Suunto has restricted this initial release of the technology to a tightly limited set of applications (read: I think it only works whilst running), but with plans to extend this technology to other sport profiles as it matures, such as outdoor swimming, which is the bane of accurate GPS tracks.

I think the 9 cements Suunto's position: they are not going after Garmin's market segment, but instead developing out their own, comprised of people who do not need a slew of features of questionable accuracy, but instead want the basics, beautifully and simply implemented. Whilst their backend is problematic at the moment, and its strategy unclear, the hardware (and firmware) on the wrist has finally solidified. (Their bright-orange wrist strap is a lot, lot better than Farmin's, too. Very important to consider.)

Finally, to anyone with a Spartan - especially a WHRB - the reasons to upgrade are as follows:

  • Shiny new toy
  • You regularly exercise further than your battery life allows you to record
  • You have a model without (a manometer|wrist HR) and want (a manometer|wrist HR)
  • Curiosity in an implementation of inertial navigation