Tuesday, July 28, 2015

Sense About Some Sensors

Sorry if this post is a sensitive subject.... (I'm terminally punny).

Today I finally broke the year long record of no outdoor runs. I've been biking to work more often lately which has been a very nice return after completely falling away from it last year and through the winter and sadly much of the spring. Now I'm cycling twenty to forty miles a week. Better than nothing. But biking to work AND running is better than that. So thats what I did today.


 The Mrs. took the twins to see a movie and neither I nor my two year old sit through two hours very well, so we opted to go for a run together. Its just under 6 miles to go to the little reservoir at the mouth of the canyon and back, adding a little back and forth in front of my house made it a full 6. I was a little nervous going that far since I don't remember when I went running last, but really I felt pretty good the whole way (I even considered extending it a few miles). I kept my slow goal of around 10 minute miles (until I hit the half-mile of 8% grade hill on the way home). I walked about a tenth of each mile (run-walk-run) so I didn't totally thrash myself. All with a stroller. So I felt pretty good.

As an experiment I had 3 different devices measure our jaunt. My wife's new Garmin 220 watch, my phone with the strava app, and my bottom-end fitness tracker made by the iFit group at work. I was curious to see how the results differed.

The two GPS devices were nearly perfectly in sync in time (1:04) and distance (6.0 mi). The minute discrepancy was probably more a product of my turning them on and off one at a time. Training for our second marathon, we tried to use my wife's new (at the time) Android 2.2 smartphone. Man those results were wild. I think her device maybe had a faulty gps module or buggy stack because it was very inaccurate. So either phones have come along nicely, or I just haven't used enough phones, but now I'd really say you can totally track distance with your phone. However I really liked the gps watch for being able to track progress, mile mark notifications, and the current pace reading. It has fewer diagnostics than the phone (and I didn't feel like uploading the data to a computer for strava), but was much nicer for the actual run. I'd have trusted its distance measurement first, but the consensus was even better.

Now for the fitness tracker: I know its just accelerometer data which drifts so much with temperature or saturation you can't ever rely on it for accurate distance tracking, but actually it wasn't as bad as I thought. Don't get me wrong, it was almost 15% over the actual distance I ran, but it could have been worse. And actually the calorie estimate was less than 1% away from the strava estimate (they are always estimates, lets not kid ourselves). The tracker said I took nearly 11,000 steps during the time I was running which calculates to a running stride of 3.2 ft (using its distance measurement). Maybe thats accurate. I don't know. Maybe I'll actually measure my stride sometime (its on the todo list). I'd love to see the algorithms various trackers use to estimate steps and distance, but I'll probably have to crack it myself if I really want to know.

So let us reiterate: just because your fitness tracker claims you burned off that doughnut, maybe take an extra lap just to be safe, (and the apple watch won't be much better, sorry). If you want to measure distance, use a GPS. If you want to measure calories you either need to measure how much total heat your body generates during exercise or how much of the oxygen you inhaled got exchanged with carbon-dioxide. Neither sounds very comfortable, so a relative value from the fitness tracker will do fine for me.

No comments: