Replacing the Sticks - How to improve football marking accuracy

Submitted by JeepinBen on December 18th, 2017 at 3:26 PM

In the wake of an index card being used to make sure that there wasn't a first down last night:

Image result for raiders cowboys first down

along with our multitude of complaints about B1G refs (JT was short), may people have come out and said "Why can't they just put a chip in the football!"

Well, as a Michigan educated mechanical engineer, let me tell you: they could, but it's not easy and it's not cheap.

The Ball

The first problem is that there can be absolutely no changes to the football itself. This is less of an issue in college than the NFL, where different universities have different sponsors and different footballs. Nike, Adidas, etc. Each might be slightly different. The NFL has one football (unless you're the Pats, ZING!) and it's gotta stay the same. In order to "track" the football, at a minimum you'd need to add sensors and power supplies. And those can't change the weight of the ball or the balance or the flight. The NBA tried to change their basketball a few years ago (to a sythetic leather at the urging of PETA and others) and it lasted less than 1 season because the players didn't like that it was different.

Image result for nba synthetic ball experiment

Good luck changing a football, as there aren't 32 people talented enough to throw them already.

Let's say that a team is able to create a weightless sensor/power supply that can also withstand all the batting around that happens to a football, a football is not a simple shape. The sensor must know where the ball is in a 3D space. Think about someone stretching the ball out forward, vs a runner who has it sideways. If the tip of the ball breaks the plane, TD. If the tip is sideline to sideline, it might not cross the line to gain.  The sensor will have to know where in the ball it is, and the ball's orientation. Again, not simple. The sensors won't be able to move even when the ball is kicked, pitched, tackled, etc.

Image result for NFL football cross section

I think this may be do-able. Put tiny sensors in each end of the football along a "spine" that goes through the center of the ball with its power supply. An inflated ball weighs just less than a pound, so there isn't much weight to work with, but it's not 100% impossible. You'd need sensors that weigh next to nothing and power supplies that fit that description as well.

The Field

On the field you'd need some kind of equipment that can precisely spot the ball and its position on the field. These sensors can't be interrupted by people, can't be line-of-sight, can't have weather issues, and can't obstruct anyone's views. They can't be in the way of players on the bench substituting, and they need to know  when the runner is "down". I have no idea how the runner being down fits into this, because that's discretionary. A system could cover things like forward progress, but you'd still be relying on a referee or a replay to show when a runner's knee touches. Which gets us back to the same problem...

Image result for 3 blind refs

We're an educated bunch here at MGoBlog, anyone have any other ideas?

Comments

rc15

December 18th, 2017 at 3:54 PM ^

The easy one to me is to require a camera to be constantly pointed down the line to gain (or goaline) from both sides. It amazes me how often they don't have that angle to overturn a call on the field.

They always talk about stitching together multiple camera angles, but they never actually do it... It shouldn't be that hard to find an exact time down to a millasecond that the knee is down on one camera, and then check another you can see the ball in at that time.

From a technology perspective, I've always thought they should be able to do something with the same cameras the NFL uses to do the 3D rotational replays. Have a sensor in each chain marker that essentially creates a virtual first down plane across the field. Then be able to rotate in 3D to see when the runner's knee is down, and then rotate to see if any part of the ball is past the first down plane.

None of these would have any technology needing to be added to the ball.

rc15

December 19th, 2017 at 7:59 AM ^

I'd add that replacing the sticks should be completely seperate from whether or not the knee is down. Refs too often round to yard lines on punts to make their job easier or give an extra 1/2 yard on a run when they think the runner got the first down to make it so they don't have to measure.

If there was a system where the ref could place the ball on the field like they do now, and get a red/green light immediately to signal first down, then they'd likely be more honest in their spots. That system could essentially be like the sensor on your garage door. Clear everyone out of the way and if the laser is broken, first down (assuming they can make the judgement fine if its not within a half ball length either way). Make it a 15 yard penalty for trying to interfere. 

ska4punkkid

December 19th, 2017 at 10:27 AM ^

I think they could make it work by using sensors on the field but not in the ball. The ref would spot the ball as they usually do and then all measurements or challenges would be done using the sensors. 

Not sure if this is even a thing but maybe some type of invisible laser grid where the laser lines only show up on a monitor if they are needing to measure or review

Vasav

December 18th, 2017 at 3:54 PM ^

Great diary. it should be doable to use sensors on the sidelines and in the ball to create a virtual ball and field that track the real ball on the real field. From there, the "chains" should be able to be set remotely by a simple controller - the "chain crew" could be reduced to one guy, or given to the replay official in the booth. Every first down the chain crew toggles a button that sets the line to gain.

The real question is - as you mention - how to determine when a runner is down? Short of putting sensors on every player all the time, I think a more elegant solution can have the chains light up when the program senses the plan has been broken - like the backboard lights up in basketball when the clock strikes 0:00. That way an official on replay can determine when the chains light up, is the runner down? Same thing for the goal line - was the runner down before the endzone lights up?

Currently, reviewing the spot fo the ball is barely more than theatrics - and since officials get "catch/no catch" wrong often enough with replay, it makes replay in general feel like another excuse for the guy in the red hat to waste our time. Having some visual indication of when the ball has actually crossed the plane can change that - make replay actually be worth a dang.

OwenGoBlue

December 18th, 2017 at 4:01 PM ^

Technology has traditionally been woven into sports slowly and cautiously so I think you start with an approach that answers the "did the runner break the plane" question on touchdowns. I'm not an engineer but it seems easy enough to place any field-level equipment needed to monitor the sensored-up ball underground and/or in/beneath the pylons. 

Prove the concept, save some time on replay/review, and refine the approach from there. Also a simpler area of the field to work with as you don't have to worry about runners giving up forward progress from the endzone.

kevin holt

December 18th, 2017 at 4:06 PM ^

Yeah sort of what the other commenters alluded to, just have the powered sensors in the chain markers (and make sure the chains don't have to get out of the way when people are running straight at them---maybe the sensor markers would be back from the field and not on the chain gang itself?)

Then that just means the ball has to have something which can be detected by the sensors which, to me, means they could be unpowered and extremely light. Someone correct me if I'm wrong---isn't this how they do it in soccer now? The ball has a weightless chip or something in the midde of it and the goalposts have sensors?

On the other hand, this solution only fixes the question of whether the ball crosses the plane for (1) a TD and (2) a first down. Maybe there can be a sensor that judges the position of the ball at any given moment so you can take the human element out of ALL ball-spotting? E.g. a sensor in each of the goalposts or pylons? Of course there needs to be a way to convey that information to the refs.

My solution: have a grid of electromagnets underneath the grass and a complementary magnet in the ball. After every down, the sensors automatically spot the ball and the officials set the ball down and turn the magnets on. The ball automatically drags itself to the exact correct spot on the field where it should be snapped from.

mfan_in_ohio

December 18th, 2017 at 5:01 PM ^

Unless it is a runner reaching for the end zone, the question isn't "Where is the ball?", the question is "Where was the ball when the runner's knee/elbow/left ass cheek hit the ground?" It's hard to find a way to put together a network of sensors that can answer that question without slow motion replay.

TruBluMich

December 18th, 2017 at 6:05 PM ^

Passive RFID solves the problem of the spot using electronics?  The inlays weight next to nothing and require no power supply. Place the RFID in the exact same spot on each ball will give you the deminsions of the ball sense thay are supposed to be uniform.  They could place the antennas along each 5 yard marker, the goal lines and around the entire field.  Then just triangulate the position of the ball based off feedback from each antenna.  Now that only addresses the spot of the balll at any given point.  There may also be a way to use antennas placed above the field.

The truly hard part is finding out when the player was down.  Which outside of sensors on the players themselves, would require human and human error to play apart in it.

It would however be useful as a true or false.  At anypoint did the tip of the ball cross the yard to gain, which could be determined on first down not by the sticks, or the goal line?

A2toGVSU

December 18th, 2017 at 9:09 PM ^

But instead of there being a central fob, somehow engineer that into a thread that is woven into the seams of the football. That solves the issue of the position and dimensions of the football. I have no idea how to solve the timing of positioning the football at the exact moment that the runner is down.

JBDaddy

December 19th, 2017 at 10:49 AM ^

Was going to say this about RFID.  Even if there were multiple sensors on the ball to give orientation, RFID's range and accuracy aren't good enough to solve this problem at present (at least, not in the loss-prevention or access-control capacities I've been shown).

Next, what if gloves start getting similar RFID-shielding materials to try confusing or obfuscating sensors to angle for "benefit of the doubt" human interpretation when the tech fails?  Do we argue then about unfair advantage or sportsmanship?  Make rules about materials?

May as well not take the humans out of the process in the first place, to me, it leaves some of the "glory" of the game (arguing over calls/spots) intact instead of making the game closer to a contest of mere precise measurements, statistics and calculation.

 

Arb lover

December 19th, 2017 at 4:21 PM ^

RFID as you all mention above with no more than four sensors around the field could kick out a space and time location for the football. As long as this is sync'd with video replay frame by frame, they could get an exact position for the football in space, for every frame (make it a glowie football or something that emits from the normal picture) so even when players are covering the ball completely, the football position would be known. Then just freeze the replay where you want and determine where the ball is. 

As a side note for you Michigan history buffs, a similar mechanical development was made for NCAA swimming results by a former Michigan physics professor and close family friend, so its more than possible and the university would likely support it. 

Clark Hopkins, a professor of classical archaeology put ‘Parky’ to the challenge of designing an automatic swimming and judging timer. At that time, many uncertainties and disputes in competitive swimming and a controversial decision in the outcome of the 100-meter men’s freestyle race in the 1960 Olympics highlighted a more universal need for an accurate timing system and re-stirred an interest in the system Parkinson had been working on in 1958 with Michigan swim coach Gus Stager. Bill accepted the challenge and in 1961, Parkinson’s invention of the electronic judging and timing system was put to trial at the Michigan high school Class A championship meet. In 1962 the National Collegiate Athletic Association approved its use ‘for all swimming events’ bringing the timing of swim meets to a new precision.

mangledpenguin

December 20th, 2017 at 2:38 PM ^

Passive RFID tech has been around for quite some time now and i was thinking along the same lines.  The power is not a factor, especially when you consider that they've put an invisble yellow line on the telivsion screen and it's been there for almost 20 years.  This is a very solveable problem if they want it to be... the thing is, they don't and that's why we have someone with an index card measuring an arbitrary placement of a ball with a error of +/- 8 inches.  This is the same league that doesn't understand any of these and are HAPPY to make fun of science.

 

sidenote: if we can sync a music video to a cat dancing, i am pretty sure we can sync a location of a football to game film.

1WhoStayed

December 18th, 2017 at 10:53 PM ^

I’ve thought of applying technology for this purpose but it just won’t make a big difference. As long as referees can throw a flag (or not) on every play - what’s the point!?

And don’t me started on “completing the process”. Or how it takes several minutes to finr irrefutable evidence to overturn a call on the field.

Just WIN and it all goes away.

truferblue22

December 19th, 2017 at 6:03 PM ^

I've always had a strong distaste for the "you should have done enough to keep it out of the refs' hands" argument (which I assume is what you meant by "just WIN"). In a sport as competitive as FBS football and a sport such as this where the officials have an incredible amount of ability to influence the game, all it takes is two or 3 bad calls to completely swing a game in the opposite direction. In a sport like basketball which is much higher scoring and has exponentially greater possessions, bad officiating is easier to overcome. In football even just one bad official can significantly influence the outcome of a game. 

blueblue

December 18th, 2017 at 11:08 PM ^

The film can be time stamped with the exact time of every frame, and the 3D positional data time stamped as well. For critical calls, replay can say exactly which frame of video the runner is down, and then you can use the rfid positioning system to precisely locate the ball at the same exact moment in time.

dyoder

December 20th, 2017 at 3:36 PM ^

Yes.

Particularly in combination with sensors, and if introduced incrementally, as several other comments have suggested about sensors. There's massive amounts of visual training data available and clear hueristics for use with sensors (has the ball moved ten yards). AI and technology in general could go a long way not only in solving the problem of determining first downs but many other otherwise difficult calls. What's more, they can start by simply providing “recommendations” with a confidence level to inform the referees, so it's not all or nothing. 

milk-n-steak

December 19th, 2017 at 9:31 AM ^

The largest crock of crap to me is that a Ref puts his foot down signalling where the ball is to be spotted - fine, when the guy was down is a human decision. 

The crock is that, if it's a first down, one of the stick guys just eyeballs the position of the ref's foot from 20+ yards away and plops the point of the stick down.  The other guy stretches out the chain from this spot. 

When we spot the starting location (yard 0) by eyeballing from the sideline, yard 10 should only ever be determined the same way.  The mere idea of measuring and saying you are 1 inch short of the first down implies the guy who determined "yard 0" was perfect which he was not.

 

Carcajou

December 19th, 2017 at 4:01 PM ^

Yes, but when you think of it, "yard 0" doesn't really matter all that much, It's OK if it's a little arbitrary, as long as it is not egregiously bad. You just start/retain possessopm from that point, with 4 more tries to advance the ball or score.
"yard 10" and the goal line are what really matter.

Carcajou

December 19th, 2017 at 10:40 AM ^

Any sensor on the ball would be least intrusive it was placed under the laces. That's where it would create fewer problems with weight distribution and the "feel" of the ball, and offer it the most protection.

The officials would be required to place the ball as perpendicular to the yard line or goal line as posible.

Carcajou

December 19th, 2017 at 10:46 AM ^

No kidding- I think that was the name of a device used in one of the NFL's league rivals years ago. I think it was a single stick to take the place of the chains.
Anybody know how it worked?

stephenrjking

December 19th, 2017 at 11:05 AM ^

I've done some thinking about this in the past.

1. A sensor needn't be difficult. If it can detect two points anywhere in the football, simple extrapolation can describe the position of the whole football if the sensor locations are known.

2. People often suggest that this is difficult because wiring players to detect when they're down is difficult. True, but also irrelevant--they are making the perfect the enemy of the good. A simple ball position sensor system fixes 80-90% of the difficulties because in most cases spots are difficult to judge not because we don't know if a player is down but because judging the relative position of the ball off of the ground is impossible due to the angles of the cameras. Even a slight offset in camera position obscures the true position of the ball. With an accurate sensor reading, computers can extrapolate the real position of the ball for every video frame available to refs. It is then a simple matter of identifying when the player is down/etc and using that ball position.

The only time things are really intractable is in mass-body situations like a QB sneak where not only the ball but the ballcarrier are completely shielded from all cameras.

Just because we can't fix every problem doesn't mean we shouldn't try to fix some of them.

3. There may be some existing limitations. Regardless, this seems like an ideal opportunity for an engineering school (say, Michigan's) to rock the world and gain prestige. Or for the NFL to use some of its millions to offer an "X Prize" to the first workable solution developed.

BursleysFinest

December 19th, 2017 at 11:22 AM ^

Didn't read all of the comments, but those RFID chips that Marathons use to track runners would be a possible solution... don't know how precise they can be, but they are near weightless.

Arb lover

December 19th, 2017 at 4:25 PM ^

You are likely considering only the current uses. Say the marathon pad or box. They are getting that position from maybe two RFID readers. If you have four (or more) you can really improve your positioning. The technology is not that much different than phone or GPS location which can nail people down pretty well, with sensors often miles away. 

Carcajou

December 19th, 2017 at 4:51 PM ^

"nail people down pretty well"- they can get close(r) with more readers, sure. But again, that's going to be a few feet, maybe even a few inches, depdingin.

But what trips the reader is when the chip gets within range. You can better triangulate with additional readers (one buried under every 5-yard stripe? Expensive to install, and it may not be possible on some surgaces), but still they wouldn't get acurate down to less than a few inches. There is also the question of supplying the readers with power. The ones I was dealing with were actually battery powered, but the range of the reader would fluctuate depending on the status of the battery.

Arb lover

December 20th, 2017 at 6:26 PM ^

That's purely a function of the technology (which should improve over time) and more importantly, the power level of the reader. Stores do a pretty decent job of tracking customer movements just having a couple readers, even larger department stores, so I think its still quite possible.

mjv

December 19th, 2017 at 11:28 AM ^

The better camera angles is the only suggestion that I read here that seems at all reasonable.

Everyone loves imagining how technological solutions can make everything marginally better, yet fail to consider the downside of when it doesn't work.  

How often have teams complained about wireless headsets not working on the sidelines of football games?  What happens when the laser grid or other alternative dies and the low tech solution has been entirely scrapped?  

Beef up instant replay with more camera angles and if possible, stitch together the angles.  If that fails, nothing has been lost from the low tech solution that generally works well. 

Spotting the ball is fall less of an issue than terribly inconsistent pass interference calls.

AC1997

December 19th, 2017 at 11:35 AM ^

Small bites.

We need to break the problems of the chain gang down into all of their components and see which ones are most important to solve and which ones can actually be solved.  You shouldn't stop trying just because you can't solve all of the problems at once.  I like the idea of the NFL using their billions to sponsor some tech projects to try to solve things.  

Here are the problems as I see them off the top of my head:

  1. Knowing where the ball was at the end of a play
  2. Making sure the "chains" are set accurately to where the ball was placed
  3. Tracking the moment when the ball carrier is down to where the ball location was
  4. Breaking the plane of the endzone

We've seen Tennis figure out how an ideal way to judge line calls, soccer has figured out the goal mouth problem, and baseball is on the verge of automated strike zones.  In my engineering career I've seen automated inspection equipment that scans components to make sure they are the right size or position and I've read enough about the defense industry that I know they're doing crazy things with tracking of objects.  

Making it perfect will be immensely difficult.  Making it better than the joke we're using now shouldn't be hard at all. 

Carcajou

December 19th, 2017 at 4:33 PM ^

Ever since they started reviewing every TD (and any close first down as well), I have thought: "before instead replay, that would have been a touchdown- they would have 'given it to him'." 

I wonder if it wouldn't be better to amend the rules to state something like: "the ball shall be marked where the ball carrier completes his [natural] fall forward to the ground."

Eng1980

December 19th, 2017 at 7:31 PM ^

I would rather tolerate a few more bad calls than ever give OSU another opportunity to get a contested, critical, close, and key first down without a review.

With all the holding and pass interference that is called and not called erroneously and inconsistently I can do without the goal line reviews.

It is really brutal when they get it wrong after review.

And please let the player with the ball finish falling forward for the mark.  I get really annoyed when they try to match the position of the ball to when the knee hits the ground.  They usually get the break the plane call correct but they rarely get the ball placement correct if they decide the ball did not break the plane.  It is as if there is a rule that if you don't break the plan the ball gets backed up feet and yards not inches.  Please finish the football play/motion.

I don't think the exact spot is key to the enjoyment of the contest.

Pepto Bismol

December 21st, 2017 at 8:50 AM ^

I have no idea how anything works, but Tennis can identify the exact path of a tennis ball in relation to the lines on the court. You see the replays a million times in a televised tennis match.

I can't imagine they have any kind of weighted equipment inside every single ball, but even if they do, it's obviously not heavy enough to affect a tennis ball.

How do they do that with such precision? So um... just do it that way.

Gene

December 21st, 2017 at 6:29 PM ^

It's pretty easy to do in tennis with cameras and machine vision because you can always get a clear view of the ball and court surface. You could do it the same way for football if you make all of the players invisible.