Yuzuru Hanyu’s thesis on Motion Capture

[treblemakerem]

Came across this video today https://youtu.be/3c2dDaIaifw

The link to his thesis is posted in the description but my Japanese is not good enough to read it. Wonder what people think about this?

 

[Blades of Passion]

I love it and want it. Please have the computer tell everyone the exact amount of rotation and amplitude on a jump so I don't have to anymore! The computer will assign GOE's better than most judges.

 

[EdgeCall]

In the last couple of years I have wondered what is going on with Yuzuru and if he is "on his way out". Looks like, despite what he has shown before, I still underestimated his determination. Having become bored with winning medals he now puts his popularity and his experience into a push to completely revolutionize the judging system and in the end the entire sport. He has my full support.

 

[gkelly]

I love it and want it. Please have the computer tell everyone the exact amount of rotation and amplitude on a jump so I don't have to anymore! The computer will assign GOE's better than most judges.

Computers will be able to do a better job of measuring the rotation, correct takeoff and landing edges, height and distance, and speed of a jump, and most errors.

But others of the positive GOE points are more qualitative, even subjective, and therefore less appropriate for computer determination:

Effortless throughout (including rhythm in a combination or sequence)
Steps into a jump, unexpected or creative entry
Very good body position from take-off to landing
Element matches the music

 

[CanadianSkaterGuy]

It's interesting, the use of motion capture to measure rotation, etc. But a couple initial issues: for things like edge and foot pressure/contact points seen in the video it seems that a skater would actually have to be "wearing" the actual motion-capture hardware in order to have the computer capture the data. The costs of this would probably only make such technology accessible for certain skaters/competitions.

Another issue I foresee is that this would have to be calibrated for every skater. On things like distance/height, a smaller skater might jump differently relative to their height compared to a taller skater. In terms of average speed, a skater with longer legs might cover more distance - or if they take a choreographic break in their program it might skew their average speed lower.

Fair point too about judging fatigue, but that is the nature of any refereed/judged sport (I don't think it takes into consideration that there are floods and warmup groups, where the judges get a bit of a reprise).

Obviously this is a fan video that is super biased and even not-so-subtly shades his rivals, which makes me question the objectivity and even the translation of the thesis itself, but it's a cool glimpse into something that may eventually be integrated into the sport.

 

[Blades of Passion]

Computers will be able to do a better job of measuring the rotation, correct takeoff and landing edges, height and distance, and speed of a jump, and most errors.

But others of the positive GOE points are more qualitative, even subjective, and therefore less appropriate for computer determination:

Effortless throughout (including rhythm in a combination or sequence)
Steps into a jump, unexpected or creative entry
Very good body position from take-off to landing
Element matches the music

Actually you could program all of that into a computer. There might be a couple details that are currently too difficult to fully describe to the computer, but competitive skating has already become robotic anyway, so it's not as if these details are being properly assessed anymore.

Also, the current guidelines for GOE are far from being the best ruleset.

Most importantly, the vast majority of judges suck. So who really cares about their hackneyed personal assessment (if it can even be called a personal assessment, given how they are essentially just puppets). The computer in nearly every case will do a better job than any human panel we've had to date.

 

[cathlen]

Some things can't be simply measured, but some can, and I'm all for it. But I predict there still would be problem, just like with VAR in football, because afterall, it's not technology that makes the decisions...

 

[kolyadafan2002]

Actually you could program all of that into a computer. There might be a couple details that are currently too difficult to fully describe to the computer, but competitive skating has already become robotic anyway, so it's not as if these details are being properly assessed anymore.

Also, the current guidelines for GOE are far from being the best ruleset.

Most importantly, the vast majority of judges suck. So who really cares about their hackneyed personal assessment (if it can even be called a personal assessment, given how they are essentially just puppets). The computer in nearly every case will do a better job than any human panel we've had to date.

I was thinking inputting accurate skaters heights and then using % of skaters height for that.

 

[lariko]

I think adding as much metrics as possible to aid technical panel’s decisions is a great idea. I think it will be particularly useful for athletes without reputation, because if metrics show an outstanding jump, it will hopefully push their GoE higher than they would have otherwise gotten. Maybe?

I think part of GoE will always still be that wow, that was the most beautiful frigging jump I had seen! but metrics are good, obviously prorated to skater’s height/stride