The new Suunto 9 features Suunto’s new FusedTrack feature.
Here’s how it works for running: GPS positions are sampled relatively infrequently (Endurance Mode 60 secs & Ultra Mode 120 secs). Between those points the Suunto 9’s motion sensor provides inputs to a directional algorithm to joint the dots BUT it does not draw a straight line. It tries to predict where you were running in between those points effectively using inertial navigation.
On the face of it, this sounds ‘a bit rubbish’. Bear with me, it’s actually quite a bit more accurate than you would think.
What could possibly go wrong?
Quite a lot could go wrong.
The GPS positions themselves are subject to many kinds of inaccuracy for many reasons. Suunto are even using a new GPS chip.
How vigorously you swing your arm might change and that might impact on the algorithm. You might be looking at your watch for a couple of seconds with your wrist not swinging at all.
You might do a sharp turn? How sharp a turn can the algorithm predict? You might perform a long, slow turn. Would the algorithm even notice that? What if you quickly double-back on yourself?
You can ponder those points for a while and think of lots of related ones for yourself. That’s what I did when I first read about it and I thought “Nah, that’s not going to work. It will just be a fancy, squiggly dot-to-dot.”
And Suunto have specifically requested that I look at FusedTrack. To which I will oblige over the coming weeks.
Then I saw DCR’s run-in-the park, below. Hmm. That’s actually not a bad track … for what it is intended to be
OK it looks like a 2k run. Something like that. And there are some comments on the forums to the same effect.
But then there is this 30k run from GZS (Source: uhrenundtouren.com clickable to google maps)
I only had an hour or so to play yesterday, so I did part of my GPS test route (details of the GPS hazards in this link to: the5krunner.com). The route has gently flowing curves, trees, flatness and some sharp turns. There are bridges, tunnels and high buildings but I’m not sure they are quite so relevent to FusedTrack.
I used the same tech setup as DCR (for no particular reason) Suunto Spartan ULTRA, Suunto 9, Garmin Forerunner 935.
When enabling the Endurance/ULTRA modes on the Suunto 9 a compass calibration is required – I did that and I even set the correct compass declination, although I guess that makes no difference to relative compass positions (please let me know if it might).
BLUE: Garmin FR935: This is recording at 1-second intervals with GPS only. Worn on the right arm on the underside. Wearing on the underside goes some way to explain the ‘alright’ track – but wearing it like that avoids damaging two adjacent watches. That’s why I did it.
GREEN: Suunto 9: This is in ULTRA mode sampling every 120 seconds and there is no GLONASS on the Suunto 9. Worn on the left arm in the correct position.
RED: Suunto Spartan Ultra: This is set to ‘OK’ GPS mode only with 60-second intermittent sampling. GLONASS was OFF. Worn on the right arm in the correct position.
So we’d expect to see a fairly good track from the 935 (because of the caveats I mentioned on the wear position) then the ULTRA will be ‘next best’ at a 60 second resolution and then the Suunto 9, effectively with beta firmware, a new GPS chip and 2 minute resolution will be the worst. Right?
Here’s what I got at a high level.
We would expect the red line from the Suunto SPARTAN ULTRA joins the dots for the position taken every minute. Naturally it cuts corners because of that and maybe also, at the top, you can see that the GPS position might not have been acquired properly over a couple of minutes sending me through the house. But the Suunto 9 track (Green) at this level looks pretty close to the blue track of the 935. VERY SURPRISING.
Here is an enlarged detail from the same route from near Eel Pie Island (that’s where the Rolling Stones started).
I’ve done this particular segment MANY, MANY times and I can tell you that the Suunto 9 is as good as some of the ‘normal’ GPS devices, admittedly at the lower end of the GPS accuracy scale. There are some tricky bits here like leafy tree cover over a road with 3m high walls either side at some points. On this section, the Suunto 9 actually does better in FusedTrack mode than on BEST mode when I tried it last week on a previous beta firmware version.
Just after the above segment, we go onto a sweeping bend along the Thames tow path (below). Very pretty. It’s also very open, except that the towpath is covered by some large and leafy trees. If anything the Suunto track is actually better than the Garmin one on this curve!! (Again the 935 was on the underside of my wrist).
Further to the top of the image, where the track crosses the bridge and does a U-turn, the Suunto 9 does ‘lose it a bit’ and, for a normal GPS test, this would be a pretty bad result. As I proceed down the right hand side towpath of the river it’s not so good there either. But this is 2 minute GPS fixes and inertial navigation!! Cool.
This final section is an easy sweeping bend on a fully open road. Super easy for GPS. After turning off there is some more tree cover before crossing the river again.
For what it is, it’s extremely impressive.
It’s VERY significantly better than I expected.
Let’s temper the excitement with a ‘So What?‘ moment.
Simplistically I’d say “FusedTrack doesn’t affect me. I never let my battery run low and I would never need to use the battery-saving modes of the Suunto 9. It is zero use for my training and zero use for my racing”
But FusedTrack is not intended for me. I can see how this will be a really nice-to-have for some of you more adventurous types off in various wildernesses for days on end.
Then I got thinking about some results I had this year with the Fitbit Versa (bear with me!!) and the Suunto 3 Fitness. I tended to use both with just their accelerometer producing the current pace. The Versa was rubbish but the 3 Fitness was very much closer to reality quite a bit of the time. So this provokes the though that maybe Suunto are using the same or similar motion-tracking algorithm? I seem to remember that Garmin also use a wrist-based accelerometer to smooth INSTANT PACE but I’ve never found that to make any important difference to PACE. I’d also previously assumed that the Suunto SPARTAN SPORT had a generally superior (but not perfect) INSTANT PACE simply because its GPS *IS* better. But maybe the Suunto has better INSTANT PACE because the FusedSpeed that produces PACE also uses the same accelerometer as for FusedTrack in the 9?
Suunto MAY, just may, have produced something special with how they handle their accelerometer.
This technology might feed through and give us better distance and pace accuracy when running. Now *THAT* I am interested in.
This could be making Apple sit up and take notice as well. The average sporty user of their Watch is probably not looking for athletic levels of precision (sorry if I generalise). but they will want a reasonable PACE figure as well as a reasonable GPS track of where they’ve been. And, for sure, they will be complaining when the battery runs out because of the GPS chip sucking the juice out of the battery. Well Suunto’s FusedTrack and FusedSpeed have gone a long way to proving to me that this sort of technology, whilst not new, can be useful for many segments of the sports market across a diverse range of sporty uses.
I think I also read that DCR was informed that a variation of FusedTrack will be used for the Suunto 9’s open water GPS track. That will be good as open water GPS was NOT one of the strengths of the SPARTAN, in my experience.
Perhaps also some tinkering with bike GPS tracks and speed sensors in forests, in cities and through tunnels could be used to produce improvments to those kind of tracks?
I’m happy now. Some tech works (or seems to…#beta). Yay! Unfortunately I have 120 miles of on-road fun tomorrow 🙁 !!! 🙂