Hawk Eye Challenge System has Arrived!

[Oldhand]

Because Hawk Eye seems to map the flight path only in 2d, I don't think the animation is meant to show the actual path. Such an animation is nicer to look at than a picture saying in or out.

The portion in bold is correct.

The Hawk-Eye system being used for badminton is a 'lite' version.
It is not the feature-rich full version that tennis and cricket uses.

The animation we are seeing now is a pre-set sequence.

Even if the shuttle drops straight down or hits the floor at a very acute angle, what we're going to see is the same animation of the shuttle approaching (and exiting).

Right now, the only calculated-and-mapped element in that sequence is the landing-spot (the end-point).

The system has errors (but you won't hear this from an official mouth)

For now, let's say the errors have much to do with these factors:
- sudden peaks in lighting when the shuttle flips 180 degrees after being hit and during the shuttle's flight (the white, shiny feathers make this worse) which 'blind' the camera sensors
- players blocking more than one camera's view of the shuttle (this is very pronounced during doubles matches)

In cricket, the path (of the ball) to be tracked is always in one well-defined area (in front of, over, behind and on the stumps) and always in one direction - from the bowler to the stumps.

In tennis, the defined area is much bigger (and so is the system being used) but the ball does not 'burn' in the light, does not reflect highlights and is easily trackable especially because it almost never gets blocked by a moving player.

Nevertheless, what is immensely worrying is something else.

In live tests, it's clear that the landing spot predicted by the system is not always true to the actual landing spot (as shown by simultaneous video monitoring). And it's not a 5mm error either - it's much much larger than that.

Unless the BWF decides to use a full suite, the Hawk-Eye system will continue to be much less reliable than honest human eyes backed up by slow-motion video replays!

 

[Oldhand]

Seriously how is this ever better than a highspeed camera's slow-mo replay? The true picture is hidden and the simpletons are brushed off with an antiquated 3D computer animation. Have never seen a cork (on a match shuttle) that's bigger than the line either

If the BWF goes for the full version, the Hawk-Eye challenges will be transformed into yet another vehicle for in-game advertising. And it's quite likely that the number of challenges will be increased to two per game, instead of two per match as it is now.

By the way, for those who are interested in such things, the court lighting at the Singapore Open 2014 was 1200 lux. That's almost twice last year's illumination level.

 

[Oldhand]

Overheard in passing:
"Whoa, they have some new thing called the Cock-Eye."

In retrospect, that seems to be true

 

[pcll99]

...

The system has errors (but you won't hear this from an official mouth)

For now, let's say the errors have much to do with these factors:
- sudden peaks in lighting when the shuttle flips 180 degrees after being hit and during the shuttle's flight (the white, shiny feathers make this worse) which 'blind' the camera sensors

...

U meant after being hit by racket and string?

If yes, why would the hit affect the accuracy of the landing spot?

Or, you meant after the shuttle hit the ground?

 

[Oldhand]

U meant after being hit by racket and string?

If yes, why would the hit after the accuracy of the landing spot?

Or, you meant after the shuttle hit the ground?

After being hit (after leaving the string-bed), the shuttle's 180-flip causes a brief surge in the intensity of the light reflected by it. That spooks the sensitive electronic sensors.

In tennis and cricket, although the tracked object (the ball) reverses direction abruptly (as in badminton), the intensity of the reflected light remains largely unchanged simply because the ball doesn't change its basic shape.

However, for the tracking system, a flipped shuttle is a sudden change in the shape of the tracked object.
And since the accuracy of the system is affected by even brief bouts of 'blindness', the rest is obvious.

 

[renbo]

SS's drop shot that was called in and challenged by LCW really looked out. I don't beleive in that system's accuracy. The video slow-mo with replays given to the croud would be as entertaining, more reliable and cheaper.
This is not good management from the BWF

 

[Oldhand]

SS's drop shot that was called in and challenged by LCW really looked out. I don't beleive in that system's accuracy. The video slow-mo with replays given to the croud would be as entertaining, more reliable and cheaper.
This is not good management from the BWF

I agree that the current system (and its implementation) leaves a lot to be desired.

However, having said that, I'd stand by that particular ruling you mentioned.
The line judge and the system ruled it in.
Lee Chong Wei said it was out.
That's two against one

 

[pcll99]

If the BWF goes for the full version, the Hawk-Eye challenges will be transformed into yet another vehicle for in-game advertising. And it's quite likely that the number of challenges will be increased to two per game, instead of two per match as it is now.

By the way, for those who are interested in such things, the court lighting at the Singapore Open 2014 was 1200 lux. That's almost twice last year's illumination level.

- Full version will cost more money obviously. But if it is more accurate and BWF (or the hosts??) can generate revenue from the hawkeye system, no one would mind.

 

[Pakito]

It is already a given that badminton is superior to tennis in terms of speed of shuttle compared to the tennis ball. In tennis, they get to challenge every 3 times per set. Assuming it's 6 sets x 3 = 18 times. Since it is alleged that the setting up of the hawk eye costs so much, use it more not only to get the money's worth, I still think a meager 2 challenges per set is just so stingy.

 

[pcll99]

If the BWF goes for the full version, the Hawk-Eye challenges will be transformed into yet another vehicle for in-game advertising. And it's quite likely that the number of challenges will be increased to two per game, instead of two per match as it is now.

By the way, for those who are interested in such things, the court lighting at the Singapore Open 2014 was 1200 lux. That's almost twice last year's illumination level.

i notice the pictures flicker quite a bit in slow motion replay. is this an issue for the hawkeye?

[video=youtube;pxXw3Zsro1Q]https://www.youtube.com/watch?v=pxXw3Zsro1Q[/video]

 

[visor]

^ That flickering is due to the background stadium lamps flickering 60 times per second due to AC current electricity, much like your home fluorescent lighting.

 

[craigandy]

i notice the pictures flicker quite a bit in slow motion replay. is this an issue for the hawkeye?

Well I can still see the shuttle clear as day

 

[pcll99]

^ That flickering is due to the background stadium lamps flickering 60 times per second due to AC current electricity, much like your home fluorescent lighting.

some countries use 60 Hz; some 50Hz.

In China, where WC2013 was held, they use 50Hz. All England uses 50Hz as well.

If the slow mo capture rate is 60 fps, i think it will be an issue, right?

 

[kwun]

It is already a given that badminton is superior to tennis in terms of speed of shuttle compared to the tennis ball. In tennis, they get to challenge every 3 times per set. Assuming it's 6 sets x 3 = 18 times. Since it is alleged that the setting up of the hawk eye costs so much, use it more not only to get the money's worth, I still think a meager 2 challenges per set is just so stingy.

they should charge the players. $50 per challenge!

 

[craigandy]

^Lol.
There is no limit on the amount of challenges, only stingy if the player keeps wasting every ones time with hopeful eyes.

 

[cobalt]

After being hit (after leaving the string-bed), the shuttle's 180-flip causes a brief surge in the intensity of the light reflected by it. That spooks the sensitive electronic sensors.

In tennis and cricket, although the tracked object (the ball) reverses direction abruptly (as in badminton), the intensity of the reflected light remains largely unchanged simply because the ball doesn't change its basic shape.

However, for the tracking system, a flipped shuttle is a sudden change in the shape of the tracked object.
And since the accuracy of the system is affected by even brief bouts of 'blindness', the rest is obvious.

From what I've seen so far, the momentary blinding does not feature in the reconstruction, really. The static cameras along the lines are the ones coming into play for the line contests. The shuttle has a consistent attitude much before it arrives anywhere close to the line. If the stationary cameras are running at 180fps, they should be able to reconstruct the touch-down very accurately.

I find it a little incredulous that Hawkeye being an English company, (badminton is played there, I believe) could not edit the system to compensate for the smaller diameter of the touch-down area and the effect of drift. If the company is serious about having its technology being adopted by BWF, surely they could have done some more work before this RC was debuted in production. Just voicing my disappointment. I suppose they may have some real reasons for the less-than-impressive performance, and I'd sure like to hear them.

 

[Oldhand]

From what I've seen so far, the momentary blinding does not feature in the reconstruction, really. The static cameras along the lines are the ones coming into play for the line contests. The shuttle has a consistent attitude much before it arrives anywhere close to the line. If the stationary cameras are running at 180fps, they should be able to reconstruct the touch-down very accurately. (...)

Um, it's not that simple.

As mentioned before, the job of the HE cameras is not to produce video.
Their job is to break down successive images to identify a set of pixels (which represent the midpoint of the base of the cork) which can then be merged and tracked by software to predict where that pixel-set will be at a time in the future.

And if the 'set of pixels' cannot be identified as soon as the smash starts, the tracking process breaks down. Normally, they have six 'cameras' of which at least three need to generate a stable point of reference before a reliable calculation can be made.

As such, the 'blinding' (bloom, spike, burst are other terms) is a big issue especially since the process for badminton has very little time to work with.

The UK and much of Asia uses 25 fps (and its multiples) for broadcast production.

If a smash from one backline were to travel at an average speed of 200 kmph and hit the other backline, it would have covered the distance of 13.4m in less than a quarter of a second.

In other words, the entire smash would show up on about 6 to 7 frames of video.
If it were a 2x slow-motion camera, it would show up on 12 to 14 frames.
If it were a 3x slow-motion camera, it would show up on 18 to 21 frames.
And so on...

(Badminton is not rich enough to use 1000 fps cameras.
I doubt if the cameras HE is now using are even 200 fps.)

Also, IMHO, not enough attention has been paid to the margin or error.

The 3.6mm error margin doesn't sound like much but, in badminton terms, that is a 15% error margin (the width of the cork is ~25mm). Such a large margin of error wouldn't be acceptable in any real-life statistical prediction scenario, much less in finance or forensics.

This margin is rather OK in tennis or football (or even snooker).
I hope they come out with a way to reduce it to 1.2mm for badminton.

 

[Oldhand]

http://www2.geog.ucl.ac.uk/~mdisney/teaching/1006/papers/collins_hawkeye.pdf

This study is six years old and is not referenced to recent advances.
It still makes for an interesting read.

 

[Littlejohn]

These shuttle speeds are all very interesting but lets not forget that a shuttle hit at 200kph will not be doing that speed when it passes the net or hits the rear of the opponents court.

This raises the questions a} is the deceleration of the shuttle speed linear
b} if not, how mathematically do you calculate its hitting speed

Out of interest does anybdy know what the actual rate of deleraration of the shuttle is, what loss of speed is there after 1m, 3m, 5m

 

[Oldhand]

These shuttle speeds are all very interesting but lets not forget that a shuttle hit at 200kph will not be doing that speed when it passes the net or hits the rear of the opponents court.

This raises the questions a} is the deceleration of the shuttle speed linear
b} if not, how mathematically do you calculate its hitting speed

Out of interest does anybdy know what the actual rate of deleraration of the shuttle is, what loss of speed is there after 1m, 3m, 5m

...which is why I used the term 'average speed'.

And it's useful to keep in mind that the HE calculations are done early on during the flight (when the shuttle is at its fastest) and NOT when it is about to hit the ground.