What do you think? You can have your say by posting below.
If you do not already have an account Click here to Register.



Edited 1 time(s). Last edit at 2023:08:23:16:48:49 by CoochieCoo.

Interesting article SW, many thanks

[24/25 adoptee: Francois van Wyk]

Anyone noticed the surprising scrum half data in the video?


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Thanks sw, interesting article.

A little context on Dugald MacDonald.

As well as "a background in private equity." Dugald has a considerable Rugby pedigree. He comes from a high quality rugby family in Cape Town. Dugald, like his father and other male members of his family was educated at Diocesan College (Bishops) a strong Western Province rugby school. He then, like his forebears went on to University of Cape Town and subsequently Oxford University where he won a blue.

Before him, his father (also Dugald) trod the familiar Bishops-UCT_Oxford path, as did uncles Donald & Colm.Dugald Snr played for the Springboks at No8 against the 1974 Lions, Donald played No8 & flanker for Scotland in the 1970s.

So perhaps no surprise he got involved in a rugby focused technology business.

PG

Interesting piece - thank you. But it'll take all the fun out of the arguments! And will the refs be getting the live feedback - will they be told if it's forward, rather than being allowed to make a judgment??

At the U20s World Championship a direct feed was made to the Television Match Official, who provided feedback to the referee.

The feed advised on:

Throw forward
Assist with decision making on forward passes by measuring the relative velocity of the ball relative to the player as it leaves their hands, so indicating whether the ball has been ‘thrown forward’ in the act of passing.

Made touch location
With an accurate location for where the ball makes touch, the Gilbert smart ball ensures the lineout is taken from the position the ball exited the pitch.

Touched in flight
Informing decision-making around whether a ball was touched in flight (e.g. partial charge down) by recognising changes in spin and trajectory caused by a touch of the ball.

Ball over try-line
Live location of the ball will determine whether the ball has reached the try-line.

Lineout throws
Provide instantaneous feedback indicating whether a lineout throw is not straight by measuring the angle of the throw from release to being touched by a player.

All worked very well and was very fast.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

So, it doesn't do what you said? Automatically flag forward passes. It does what I told you was its limit, give some rather limited information to the TMO who can easily make that call anyway.

It does was I wrote in the post above. It will advise the TMO (automatically) who uses the direct feed to advise the ref if the ball is thrown forward it’s up to the ref to decide.

Same for the other situations described above.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Quote:

DanWiley
So, it doesn't do what you said? Automatically flag forward passes. It does what I told you was its limit, give some rather limited information to the TMO who can easily make that call anyway.

How was the wrong side of the bed this morning, Dan?

... IMHO, of course.

Now in Honolulu

Where does it say it does this automatically?

Quote:

shipwrecked
At the U20s World Championship a direct feed was made to the Television Match Official, who provided feedback to the referee.
The feed advised on:

Throw forward
Assist with decision making on forward passes by measuring the relative velocity of the ball relative to the player as it leaves their hands, so indicating whether the ball has been ‘thrown forward’ in the act of passing.

Made touch location
With an accurate location for where the ball makes touch, the Gilbert smart ball ensures the lineout is taken from the position the ball exited the pitch.

Touched in flight
Informing decision-making around whether a ball was touched in flight (e.g. partial charge down) by recognising changes in spin and trajectory caused by a touch of the ball.

Ball over try-line
Live location of the ball will determine whether the ball has reached the try-line.

Lineout throws
Provide instantaneous feedback indicating whether a lineout throw is not straight by measuring the angle of the throw from release to being touched by a player.

All worked very well and was very fast.

Thanks for highlighting this SW.

Particularly like the use with lineout throws and ball over the try line.

The thing about the line not straight one, which I think is technically quite easy, is do you really think refs are letting that go because they don't notice it? The TMOs surely do already.

At all levels of the hand I've heard refs say "look, conditions are bad, we don't want a scrum after every line out, so I'm giving you the benefit."

I think that's what bothers me about all this, the best games of rugby are those reffed by humans with understanding. The worst (for player and viewer) are those reffed by a human trying to be a machine and apply all laws in all circumstances.

I'm not sure the ball over the line helps much? You've still got to know if there was a hand underneath it and it's a pretty fine call in any case. Ball touches a blade of white grass, try. If it doesn't, no try. With an error in the order of cm is that going to give you a definitive answer?

I’m wondering how they will judge lineout not straight. When the ball leaves the hookers hand straight and after 5m the wind blows it to his side of the lineout and straight in to the scrum half hands at 20m

At present RB, some hookers might as well be throwing it straight to their scrum half!

Season before last it was commonplace for the defending side to be penalised for jumping across when the throw was coming in at a 45 degree angle towards the throwing team's 9.

That has since been improved after some DORs complained, but any scientific assistance to help reduce human error should be welcome.

PG

Again, if it's so obvious and this is an area without rugby are trying to improve, do you really think it's all down to the refs (and tmo) missing it? As you say, it's pretty obvious.

We can argue we don’t need the accuracy the smart ball is able to deliver but that ignores the whole raft of other features the smart ball can offer.

In my view it’s obvious that the algorithm can operate within agreed tolerances to aid refereeing and the fact that those tolerances are defined offers the one factor that nearly everyone agrees is essential for a good referee and that is consistency.

If we expand our horizons we see that the smart ball can provide a greatly enhanced experience for the rugby fan.

For example much has been made of George Ford’s spiral bombs in relation to hang time and difficulty of catching the rapidly spinning ball.
The data from the smart ball can analyse and assess the details of the kick, spin rate, distance, direction and hang time so that commentators, pundits, coaches can judge if those kicks are exclusive to that player.
Pass length, speed and height can be measured and displayed as a screen graphic so that we can appreciate a back skill that is often overlooked.

You would have to think that coaches have been waiting for such data for a while.
As mentioned above one thing that we need to remember is that the real rugby fan, the one standing in the rain at the ground every weekend, is actually going to be the one to miss out on this data revolution. I’m not sure how you solve that.





Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Quote:

shipwrecked
We can argue we don’t need the accuracy the smart ball is able to deliver but that ignores the whole raft of other features the smart ball can offer.
In my view it’s obvious that the algorithm can operate within agreed tolerances to aid refereeing and the fact that those tolerances are defined offers the one factor that nearly everyone agrees is essential for a good referee and that is consistency.

If we expand our horizons we see that the smart ball can provide a greatly enhanced experience for the rugby fan.

For example much has been made of George Ford’s spiral bombs in relation to hang time and difficulty of catching the rapidly spinning ball.
The data from the smart ball can analyse and assess the details of the kick, spin rate, distance, direction and hang time so that commentators, pundits, coaches can judge if those kicks are exclusive to that player.
Pass length, speed and height can be measured and displayed as a screen graphic so that we can appreciate a back skill that is often overlooked.

You would have to think that coaches have been waiting for such data for a while.
As mentioned above one thing that we need to remember is that the real rugby fan, the one standing in the rain at the ground every weekend, is actually going to be the one to miss out on this data revolution. I’m not sure how you solve that.

TBH I don't go to a rugby match to analyse the minutiae of the spin rate of the ball. I go to enjoy the visceral pleasure of watching an unpredictable contest and to shout "C'mon you Baaaarrrth" as loudly as I can in the quiet bits. Also tbh this level of accuracy is likely to detract from my pleasure as you know that the calls are probably going to be correct and so the inadequacy of the ref or the cheating of the opposition (usually the Tigs) can't be blamed for our failings.

Rugby is not a game of accuracy, it is a game of opinions (most of which are either partial or just plain wrong). That's why it's so much fun

... IMHO, of course.

Now in Honolulu



Edited 1 time(s). Last edit at 2023:08:04:06:59:54 by joethefanatic.

I don't disagree it will provide a lot of data. Will it be interesting to fans? Maybe, if the rpm of George's kicks floats your boat then great. I suspect this isn't the sort of thing they'll be sharing much, every now and again we'll get some distracting titbits like "we've just reached over 3km in total distance kicked.' That's no bad thing and it's really what this ball is all about, putting some stats on the TV.

I'm not sure it's that useful for coaching. If the little gimmick in the video about distances of passes and kicks were of interest to coaches they'd be doing it with some advanced bit of kick like a tape measure and some cones. With regard to judging the hang time of a kick chase for example I think they'd be better off getting the kickers and chasers together and, you know, practicing.

I don't think it's at all obvious that a tolerance of cm is of that much use to a tmo who can watch the thing in slow motion from a number of different angles. Being consistently strict is a terrible way of reffing a rugby game.

During Covid lockdown, NZ had their own domestic Super Rugby competition- Super Rugby Aotearoa. Referees were instructed to referee strictly to see what imact that would have- especially offside & ruck offences.

The outcome? First 2 rounds heavily disrupted play, multiple penalties & stop-start game. Thereafter, once players had adjusted, more fluent play, more ball in play time, more open rugby.

Strict observance of laws, managed sensibly should AID entertaining rugby, not hamper it.

Similar benefits can be gained with the help of smart ball technology.

PG

I agree with this. Smart ball has been used at more than one tournament recently all the trials were successful not only in terms of functioning but by how smoothly they ran.
The u20 World championship had very few interruptions of play and few 'not straight' line outs. Forward passes were rare particularly the close quarter ones.
If it becomes inevitable that pushing the laws to the limit will be exposed then perhaps you simply don't risk it.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Quote:

P G Tips
During Covid lockdown, NZ had their own domestic Super Rugby competition- Super Rugby Aotearoa. Referees were instructed to referee strictly to see what imact that would have- especially offside & ruck offences.
The outcome? First 2 rounds heavily disrupted play, multiple penalties & stop-start game. Thereafter, once players had adjusted, more fluent play, more ball in play time, more open rugby.

Strict observance of laws, managed sensibly should AID entertaining rugby, not hamper it.

Similar benefits can be gained with the help of smart ball technology.

PG

Yes the problem is consistency

Quote:

Shipwrekced
The u20 World championship had very few interruptions of play and few 'not straight' line outs. Forward passes were rare particularly the close quarter ones.

The fact you got fewer incidents indicates that either: the ball doesn't work that well and refs relied on it doing so or the effect was entirely phycological and nothing to do with the actual technology. For the ball to be effective it needs to flag more of these events, initially at any rate. To say the ball achieved this effect seems to be entirely the emperors new clothes.

I do wish they'd police the breakdown better, this ball doesn't help there. I'm not convinced being strict is the way to go, I can't believe we're even close to having a technology that sees everything at a ruck, there's going to be an intuitive element to judging what's going on for a long while yet and in many ways being strict is more arbitrary, than not. That's not to say there aren't things I feel WR should clamp down on that would improve the game, attackers for some reason being under no obligation to stay on their feet it seems for example. That doesn't mean I want a ref to have no choice in applying the law, they are laws intended for interpretation at the end of the day.

Technology has always focused on doing the donkey work quickly, allowing the person to refocus on more important decisions.

The refs/linesmen will start to trust the technology and will reposition themselves on the pitch or run different lines to focus on other areas of the game.

The data is a nice freebie that the betting companies can use for in game betting.

Quote:

DanWiley
So, it doesn't do what you said? Automatically flag forward passes. It does what I told you was its limit, give some rather limited information to the TMO who can easily make that call anyway.

Bizarre response. It does do exactly what we all discussed before. I'm sure that they could install a red light in the TMOs office which emits a satisfying 'ping' if the relative velocity of the ball is higher than that of the player as it leaves their hands. Could be a good Raspberry Pi project for a school kid?

[24/25 adoptee: Francois van Wyk]

But it doesn't claim to do that as far as I can see. I've still not seen anyone explain how the ball or the system knows when it leaves the players hands. That absolutely crucial. Forward movement relative to the player before the ball is released is fine, indeed common when you're winding up for a pass. Forward movement is fine after the ball is released. So it really is that crucial, split second as the ball is released it needs to look at and with a fidelity of hundreds of ms and cm I don't see how it can isolate that moment even if it could detect the release.

So I don't see how it can do it and I don't see where its claiming to be able to? From what I can see it claims to be able to provide additional information to the TMO when they see a potential forward pass, no to be able to detect one.

Relax DW, its literally the point of the tech, invented by all of those above - the top companies and the top developers with rugby interest. I don't even get why you are upset by this one tiny point, it detects the data and whether they choose to look at it or alerts them is really the main concern?

I would guess that most of it is PR - they don't want to say that the tech will drive the decision but really it will for binary matters such as where touch was made, which I think will be hugely useful personally.

My only technical question would be if someone really wangs it out of their hand with a massive spin - its not that uncommon for the ball to move a tiny bit forward before spinning well back by the time the catcher receives it?

[24/25 adoptee: Francois van Wyk]

As soon as a ball is released/passed, the vector of its motion will change. The sample rate of data is 20 times a second.

The data models will indicate a forward pass with certain levels of tolerance, it's the up to the governing body do settle on the level of tolerance they're comfortable with - analogous to the 'Umpire's Call' in cricket, rather than the in/out of VAR on goals or line calls in tennis.

I not really upset, you'll have noticed I like a good "discussion. " I guess I'm particularly interested because localisation of things, in various ways, is a huge part of my career and there's a subtle but significant difference between what people are claiming it can do and what the state of the art can actually do.

I don't think this was invented for refs, I think its all about the stats for the tv audience.

Why not both?

Quote:

hasta
As soon as a ball is released/passed, the vector of its motion will change.

Why? The motion of the hands will be almost exactly the same as the ball. Air resistance will marginally slow it, but distinguishing between the very point of release and other prior hand motion seems very hard to do.

20 times a second seems a very slow sample rate to me.

Because while there is a motion of the hands, the ball is also being propelled forward by the player running with it. That forward propulsion vector ends once the ball is released and it immediately starts to decrease in velocity due to air resistance. There will be a notable change in any chart.

Try counting to 20 in 1 second if you think it's too slow a sample rate. Looking over multiple data points to see the velocity and direction vectors will clearly indicate point of release - it's is not about just comparing any two points of data.

The Smart ball samples its position on the field in 3 dimensions including measuring the acceleration and rotation. It does that 20 times a second or every 0.05 of a second without fail. (Unless the battery is flat!)

The algorithm processes that data, the algorithm takes a scenario and responds to the information is being given. It can be made to decide early say after 10 data points (1/2 a second) or late after 40 data points (2 seconds). What has been decided I know not but tailoring the algorithm would accommodate Matt's concern.

The whole point of Smart ball as mentioned in the article is that it does not rely on external observation but is internally monitoring the position of the ball. It can be as precise or imprecise as you want it to be but it will always be virtually instantaneous (+ or - 0.05 sec) and always consistent. Consistency is perhaps the most important aspect. How humans use this info is another matter entirely.

It's up to the sport how it uses the data and this has clearly been agreed with World Rugby. To suggest that Sportable, Sage, Gilbert and World Rugby have not thought this through is just not credible.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Quote:

hasta
There will be a notable change in any chart.

I'm not sure about that. There's quite a lot going on there: wind could accelerate it, gravity would accelerate it faster or slower depending on how its thrown (flat or up). It's quite likely a spinning, possibly tumbling, single sensor. As shipwrecked points out you need a fair amount of data points, spread over seconds, to ascertain the general motion of the ball, detecting a marginal slow down due to air resistance to pin point a pretty precise moment of time in a very dynamic situation seems rather difficult given that.

Quote:

Shipwrecked
To suggest that Sportable, Sage, Gilbert and World Rugby have not thought this through is just not credible.

As I say, the thing that sells this is the stats to viewers, that's where the money is in this. There's no money in detecting law breeches that are pretty obvious in any case.

Quote:

DanWiley

Quote:

hasta
There will be a notable change in any chart.

I'm not sure about that. There's quite a lot going on there: wind could accelerate it, gravity would accelerate it faster or slower depending on how its thrown (flat or up). It's quite likely a spinning, possibly tumbling, single sensor. As shipwrecked points out you need a fair amount of data points, spread over seconds, to ascertain the general motion of the ball, detecting a marginal slow down due to air resistance to pin point a pretty precise moment of time in a very dynamic situation seems rather difficult given that.

It's frustrating that you don't understand this Dan in no way did I say "you need a fair amount of data points, spread over seconds, to ascertain the general motion of the ball"

Hasta has patiently already pointed out you actually need very few because you are getting 20 of them every second. There is a point at which there is little point in explaining further if understanding is not moving forward.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Yes agreed that its great for the TV visuals, they can basically VR any play. Also agree that the game isn't ruined without it (such as local rugby). I welcome it as an interesting development / bit of tech and look forward to seeing what changes (if any) it can make to the game.


As I said before I think that the most useful aspects will be whether the ball is grounded in a heap of players (assuming that it can actually manage this) and the accurate touch line calls.

[24/25 adoptee: Francois van Wyk]

Honestly I think this in one of those things that if you don't get conceptually it's hard to understand. I think if we could show the charts lined up with video it would be clearer.

"The algorithm processes that data, the algorithm takes a scenario and responds to the information is being given. It can be made to decide early say after 10 data points (1/2 a second) or late after 40 data points (2 seconds)."

is very, very close to saying:

"you need a fair amount of data points, spread over seconds, to ascertain the general motion of the ball"

Quote:

DanWiley
"The algorithm processes that data, the algorithm takes a scenario and responds to the information is being given. It can be made to decide early say after 10 data points (1/2 a second) or late after 40 data points (2 seconds)."
is very, very close to saying:

"you need a fair amount of data points, spread over seconds, to ascertain the general motion of the ball"

It is nothing like that in any way shape or form that is a spurious inference. Each point has 5 aspects after at 1/2 a second you have 50 plots. More than enough.

See Hasta's post above.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

I don't see how you're saying the two statements are substantially different? You've literally said you need 10-40 data points (or even 50-200 aspects) and they are taken over 1/2 to 2 seconds. How is that in any way different from what I've summarised you as saying?

So, can you be precise in what you're saying? You're saying now you have 50 data points collected in half a second, what are they? Location, orientation? What else? What is it more than enough for? How are they used?

Quote:

DanWiley
I don't see how you're saying the two statements are substantially different? You've literally said you need 10-40 data points (or even 50-200 aspects) and they are taken over 1/2 to 2 seconds. How is that in any way different from what I've summarised you as saying?
So, can you be precise in what you're saying? You're saying now you have 50 data points collected in half a second, what are they? Location, orientation? What else? What is it more than enough for? How are they used?

DW you can research all of this if you are (as it seems) really interested in the minutiae of the tech - SW was simply writing (what I took to be) an interesting artlicle which kept the website live and gave us a summary of what was currently happening in the area of smart balls? Unless I have misunderstood SW isn't part of the development team for this product?

[24/25 adoptee: Francois van Wyk]

Dan read what I said please.

Quote:

Shipwrecked
It can be made to decide early say after 10 data points (1/2 a second) or late after 40 data points (2 seconds).

We are referring to the algorithm. A trend is easily available at 1/2 a second even before that but if you want to average it out over a longer time you can do so. Depends on the sport NOT the Smart ball.

Each point is 2D on the pitch plus height, then add in acceleration and rotation. 5 points per sample.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Point taken Matt, part of the problem here is I do research this, its (part of) my job. From my point of view a lot of what's being said sounds like marketing rather than engineering.

Quote:

DanWiley
Point taken Matt, part of the problem here is I do research this, its (part of) my job. From my point of view a lot of what's being said sounds like marketing rather than engineering.

Could well be Dan, I'm sure that they are doing it to sell their product after all. Fundamentally though they are only going to sell them if they actually do work. These full / final details will all become available at that point I guess.

[24/25 adoptee: Francois van Wyk]

I don't know, I think they'll make their impact telling us (viewer) how far or high the balls been kicked and other bits of trivia. If I understand it right the TMO aspect of this won't be under public view, my feeling is that it will quietly be ignored, I don't think a professional TMO really needs this data to make an accurate decision. A pro refs get this right in real time almost all the time. But it may be in year to come we see them same data as they do, that will add more heat than light I suspect. What happens when the data says one thing but the TMO says another? Can you imagine the screaming at the TV from the offended side? Particularly when its one of those cases where the ball went backwards from the player but forwards on the pitch?

Sorry, I'm going on a bit.

Going on a bit ....... :-)
My take is this is only the start, getting data to the pitch side.
The trick will be the data processing with multiple balls on the pitch and the software knowing who is in possession, will probably need human interaction. The forward pass will be a good one to detect, line out throws will need more tolerance.
Set thresholds through trial and error and when proven let the TMOs sit back look for the big red warning light.
It does feel a bit like doing something for the sake of it.

There is only one ball on the pitch BB. Multiple balls around the ground but only one within the sensor area.

Its already been used successfully, there is clearly a promotional aspect but I don't think there is a market for full time kick data for example. There is a market I feel to analyse an exceptional piece of play such as a long fast pass or as previously mentioned highlighting the different spin imparted to a spiral bomb by a player and if it even exists.

This data for coaches would be extremely helpful, you can quantify passing ability for the first time. Ruck speed can be measured repeatedly and accurately without TV coverage.

I've mentioned it before but where it is important on infringements I feel is in its consistency. E.g. on forward passes you can set the algorithm to permit passes 1 degree forward but not 3 degrees forward. I suspect that already exists for line out given what I've observed.

The massive difference as I mentioned before is that it does not rely on observation of the ball because the chip is embedded in the ball.

It can be extended to indicate if a kick has gone over or in touch, or if ever it became a thing whether a put into the scrum was straight.

I don't see it as gimmick nor do I see it fulfilling only one function. It think it will be used for a combination of infringement management, data presentation to fans and commentators and coaching aids.

Once bought costs would be relatively low to run so I think it would be widely available to many.

If you look at the data that flies around in first class cricket I think it's just rugby catching up and being brought up to date.

Biggest problem is an awful lot of rugby fans don't want to modernise and are hugely resistant to change.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

Ultimately if it's no good or of no value to TV then it will disappear won't it? I don't see it as anything to get worked up about and also don't see it changing the game.

[24/25 adoptee: Francois van Wyk]

From TV perhaps, not from the field though. Coaches love data.
You could track whole back moves from lineout or scrum through a multiple phases measuring ruck speed then the angle of the backs run, the depth of the pass, its speed and trajectory to crossing the line.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

I see Sage claim it is cost effective but I cannot find anywhere what the costs will be to clubs, RFU and fans.

Sportable's website is under development, Iguess costs get revealed when you make a commercial enquiry. Hawkeye costs between £50-60,000 per court in tennis if that helps. Bigger area but only need one.

As it's portable it's probably available for hire.


Tyler Offiah
Opinion is based on assumptions, whereas fact is based on observation or research.

I believe Bath already use Sportable data from the tracker monitors worn by players.

PG

Hope it isn't tech produced by certain overseas industries. We don't want to be hacked during a game!

B4B
The provenance of the company is explained in the article and my post 13.41 on 3 Aug.

PG

Quote:

P G Tips
B4B
The provenance of the company is explained in the article and my post 13.41 on 3 Aug.

PG

Only joking PG, though it might explain said nation winning the 2027 RWC final even though our eyes tell us they didn't.

Quote:

B4thB4ck

Quote:

P G Tips
B4B
The provenance of the company is explained in the article and my post 13.41 on 3 Aug.

PG

Only joking PG, though it might explain said nation winning the 2027 RWC final even though our eyes tell us they didn't.

Have you got Friday's lottery numbers?