Posted by editor
on February 23, 2013 at 5:10 AM PST
The first regular conference session I attended at Jfokus 2013 was Christer Norström's Internet of Sports (or How to win GOLD medals in the next Olympics). This session followed the keynote address for the Jfokus Embedded 2013 conference within a conference. I've long been interested in statistics and data analysis in sports, going back to...
The first regular conference session I attended at Jfokus 2013 was Christer Norström's Internet of Sports (or How to win GOLD medals in the next Olympics) . This session followed the keynote address for the Jfokus Embedded 2013 conference within a conference.
I've long been interested in statistics and data analysis in sports, going back to my teen years when baseball manager Earl Weaver amazed me by making in-game decisions based on hand-collected statistics on career batting performance versus individual pitchers. Fast forward several decades, and the notion that embedded devices might somehow find a role in sports today did not shock me. I was quite curious, though, about what Christer was up to!
I've always loved graphs (making them based on coherent analysis of data collected from various instruments has been a central aspect of my programming career); so, when I saw Christer putting lots of plots onto the screen and discussing their import, I knew this was a session I was really going to enjoy. This particular plot was made by attaching sensors (accelerometers) to cross country skiers. The accelerometers measure acceleration (which is the second differential of position, if you remember your Physics) in three different directions (left-right, forward-backward, up-down). The data is read by a smartphone, which transmits it to a server that stores the data, after which the data can be made into plots, fed into data analysis algorithms, etc. So, the "Gear 2" plots are showing overlaid sets of movement in different directions for a specific type of cross-country stroke. And, the plot shows the motions for three different skiers. As you can see, the patterns in each column are similar for all three skiers -- similar, but obviously also different. Each skier has unique motion characteristics as they perform the same sking stroke/gear.
Christer says that you can characterize individual skiers using their three-dimensional movement patterns across the major strokes/gears that make up cross-country skiing -- creating what he calls a "Movement DNA" for each skier. So, looking at the "Movement DNA" plot, we see that if Christer was presented with a new plot of movement patterns, and he knew that the skier was either Jennie or Magdalene, he'd likely be able to identify which skier had created the new plot simply by comparison of its patterns with the known "Movement DNA" that is characteristic of the two skiers.
"So, who cares?" you say. Well, suppose that you've recorded the Movement DNA of dozens or hundreds of skiers. Suppose those recordings include the best skiers on your team, along with other skiers who aren't yet quite the best, and more junior skiers, etc. Then, you can compare the Movement DNA of your very best skiers with the Movement DNA of those skiers who aren't yet quite as good, and look for differences in their movement patterns. Perhaps the less good skier's movement pattern is in some way less efficient, or more wasteful of energy?
How about this possibility. Skier A was great, among the best in the world, two years ago. But now, he or she has fallen into the ranks of the merely very good. If you had that skier's Movement DNA from two years ago, you could get their current Movement DNA, compare the data, and try to evaluate if something has changed with respect to the patterns of movement. It's not guaranteed that you'd find a change, but it's a very interesting new tool in the Olympic team's arsenal, right?
This next set of plots graphs acceleration versus time while the skier is double-poling [Note: possibly it's force, not acceleration -- but, since f=ma, the one is proportional to the other, so my points that follow should be valid in either case]. Blue is left-right acceleration; red is up-down (the skier goes up and down as part of thrusting with both poles); green is forward-backward, primarily the thrust of the poles pushing backward to propel the skier forward.
The first plot is the data from a top skier. Christer highlights the rapid acceleration as the pressure on the pole quickly reaches maximum force (the green line's periodic rapid descent); but, just as important is the fact that the peak level is consistently maintained for a relatively long duration (notice how the green line stays quite close to its minimum value for a long time -- the pressure is maintained by this expert skier).
The next plot shows the same data for a junior skier. Looking at the green line here, we see that the junior skier does a pretty good job in the initial thrust. But, in every case, the pressure is not maintained, it's like the pole pushes back after the initial thrust. Hence, the skier is initially propelled, but then the pushing in some cases almost immediately reverts to zero, before the skier's second effort push takes place (but the second push is weaker than the initial push). What might this mean? That the junior skier lacks the strength to maintain the pressure of the initial push? Christer suggested this possibility.
Christer also compared the blue lines, the left-right motions of the two skiers. Note how occasionally the junior skier exhibits pretty extreme left-right motion/acceleration compared with the expert skier. Significant left-right motion is wasted effort, since the prime objective is to move forward as fast as possible. Is this sideways motion due to the junior skier sometimes losing balance? Or is it simply that his/her technique is just not as compact and efficient as that of the expert skier?
Christer puts this into context in a figure that illustrates the entire system, from sensors to the smartphones that read the data from the sensors and send it to a data center in the cloud, where the data can be analyzed, applied to laboratory training, potentially published in near-real time, etc. He also talked about this effort as being simply an example of the type of applications that are possible through use of fairly inexpensive sensors that feed data to smart phones which relay the data to data centers. Examples include monitoring health indices of people in their homes, etc.
The next Winter Olympics are less than a year away, in Sochi, Russia. After attending Christer's session, I'll surely be following Sweden's performance in the cross-country skiing events!
You can download Christer's slide deck if you'd like to see more very interesting "Internet of Sports (or How to win GOLD medals in the next Olympics)" session content.
Subscriptions and Archives: You can subscribe to this blog using the java.net Editor's Blog Feed . You can also subscribe to the Java Today RSS feed and the java.net blogs feed .
-- Kevin Farnham (@kevin_farnham )