Daniel Simu - #
Juggling tracking wristband
I was recently, just for fun, looking at the possibilities of fitness tracking wristbands. They measure steps, stairs, sleep data, etc. I assume with motion sensors.
Do you think it would be possible or useful to have juggling tracking wristbands?
With the acceleration data & frequency it might be possible to deduce the amount of balls you are juggling, at least for basic patterns, and keep track of how many catches are done before you drop.
All this data could be graphed, so you can see over time if your average runs increase for example.
I think that large juggling data could be very useful for any training program. When are you most effectivee (time of day, part of training, type of warm up), how fast is your progress, how reliable or stable are the tricks you think are solid, how do you progress compared to other jugglers, what is the effect of a different prop type and how long does it take to adapt to that... All questions which can be answered if there was data!
I can see that retrieving juggling data from arm motion data can be tricky, and its application might be limited to only basic pattern juggling, but do you guys think it might be possible? Are there any hackable wristbands on the market?
should be very useful for solidifying any periodic patterns mainly & especially for endurance, - but why not if you can do several runs for the software to have sth to compare - maybe for learning, even. (good idea!)
- would reasonably need two wristbands.
- should(?) be hightec (high-precision) in order to calculate heights & angles well from acceleration & vector data and be able to relate it to the juggled pattern? That would make it expensive and ..
- wonder if any manufacturer would wanna launch such juggling accessoire.
- training methods on high level could be brought to way higher perfection - I'm not sure if juggling is broadly serious competitive sports enough for such to find a spreading. For many jugglers working on endurance and precision (Isolation-effects) the time is long ripe for graphically rendered statistics - think of it combined with Ken Matsuoka's matchstick manikin, resp. JugglingLab-syntax, resp. jongl's graphics!?
- Juggling theory is ripe for such things, I reckon.
- (highspeed) video-analysis might be uncmoparably more helpful, still then, yet.
Here's an approach to 5b casc statistics - alas! in german, but the diagrams and images give an impression of the level of research - https://jonglieren.at/theoprax/wurfgenauigkeit/index.html for ("Wurfhöhe"=height) height / landing speed & momentum ("Abb.2" = fig.2); .. for ("richtiger Winkel" = the right angle) angle / precision loss ("Abb.3"); height / angle ("Abb.4"); .. comparision precision of angle 3b to 5b ("Abb.5"); .. ("Handgeschwindigkeit " = handspeed) handspeed for 3b casc, 4b fountain, 5b casc compared ("Abb.6"); .. plus some cam-recorded plot-dot-spot-diagrams on behalf of precision.
I mentionned that only for comparision with what lots of data would be available from wristbands.
- and still you'll always need someone to interpret well the data furnished and to draw the right conclusions of it. And that's aeons more complex compared to recording swimstrokes or steps jogged with what heartbeats.
(I myself wouldn't buy or use one, as I'm not so much into endurance or much precision more than comes from frequent practise.)
I can imagine a passing pattern training wristband!
It would come in a set of four, and would give a small BUZZ on the appropriate wrist just before a pass, with some kind of code to indicate what kind of pass it should be.
It should have a kind of DDR mode that makes you follow a sequence of random undisclosed passing patterns until you fall over.
That reminds me of a passing pattern that I did recently where every prime numbered beat was a pass, every composite number a self. I think we did it where we went up to about 43 before countig back down (1,2,...,41,42,43,42,41,...,2,1,2...). And we dis it where one passes straight, the other crossed (just to make it more difficult).
All counting was done silently.
Brook Roberts - - Parent #
When you get back down to 1, do 1 twice to increase the period.
I can't help but think that the period isn't the important thing in that pattern. Forty-something is enough that you don't remember the pattern (though maybe remember a few pitfalls), so the pattern is basically being calculated on the fly in your head all the time.
With a crosser and a straight, would an extra 1 swap the hands for both people on the next round?
Next time I'll try both variants, but I'm not sure that they'll feel so different.
Brook Roberts - - Parent #
I would like this so much!
If only I could be bothered to make one...
A much easier version could be make with it just calling out passes (at a regular rhythm, that you had to keep to), so that you didn't have to process any outputs. Would be almost as good, presuming you happy to police yourself.
Mike Moore - - Parent #
Possible, yes: a friend of mine had a smartwatch with an accelerometer and gyroscope, and he wrote some code to plot these variables against time for use with tracking juggling. A sample plot of my juggling: https://imgur.com/Akdp52s
Some limitations right now: The watch's reference frames are set (if you don't put it on perfectly consistently, you don't get perfectly consistent results) he is super busy so isn't rapdily developing it, it would cost a bunch (but uses standard wearable tech, so if you own a smartwatch, you could probably use it).
AFAIK, he was using a Android smartwatch, I don't know what kind.
If these ever get prevalent, you could have a game at a fest: 5b cascade for a minute with the least motion possible. Or, the obviously superior, most motion.
Any chance of a file containing the data from which the plots come?
It looks interesting, and I'd like to see what else can be drawn from the same set.
(and scraping a graph to get its values seems rather unnecessarily difficult ;))
Mike Moore - - Parent #
I don't actually have the data file, but I've messaged the person who does. Like I said, he's a super busy guy, but I'd expect a response within a couple days.
Mike Moore - - Parent #
Hah, here we are!
https://github.com/tdresser/juggle_stats/blob/master/post_processing/sample.csv
I'm new to github, but hopefully that's what you're looking for.
Very nice regular pattern but I'm having a hard time interpereting it. It looks like all the accelerations are negative but, since the motion is cyclical, I expected there would be more or less equal accelerations in both directions, or the time intregals in each direction would be the same. I tried it with my iPad for one cycle and there are accelerations in both directions that look much more equal than the ones in your plots.
Kickstart suggests other people are intersted in accelerometer data. I didn't see any juggler propsals but I didn't look at them all.
Mike Moore - - Parent #
"Very nice regular pattern but I'm having a hard time interpereting it. It looks like all the accelerations are negative but, since the motion is cyclical, I expected there would be more or less equal accelerations in both directions, or the time intregals in each direction would be the same."
You'd think so, wouldn't you! I don't know why it's shifted.
Daniel Simu - - Parent #
Cool!
And yeah, calibration would be a huge issue, wrist bands move so easily..... :(
Cost a bunch... what? A bunch of time to develop? A bunch of money to... be made by a professional?
Mike Moore - - Parent #
The plots from above were from a smartwatch, which actually didn't shift noticably while juggling.
It would cost a bunch to buy the smartwatch! But that's really like saying an iPhone app would cost a bunch because iPhones are expensive. Most jugglers probably wouldn't buy the watch just for the juggling analysis app.
Daniel Simu - - Parent #
Hmm, fitness bands go cheaper, like fitbit flex starts at 75€, it has an API but I am not sure if the right data can be extracted from it..
Subscribe to this forum via RSS
1 article per branch
1 article per post