Tuesday, November 1, 2016

First world solutions to problems you just have to have – how one man’s obsession led him a meandering path. But it’s about the journey....

Locating Tennis Balls with RFID

If you wanted to build a device, a robot device, not much larger than a robot vacuum cleaner, that would collect and return tennis balls, how would you do this?

The ‘Hawk-Eye’ system, familiar to watchers of professional sports, does an excellent job of using video imagery and statistics to calculate a best guess location for a tennis ball in relation to the boundary lines. 
from UC Berkeley
And more than once, university students have built bumbling robots that rely on a variety of remote and local video, computing and robotic systems, to find and retrieve balls. At a rate of about one a semester.

Could a video system, some modern software and a radio controlled (‘RC’) car be fused together with some odds and ends into a better than bumbling system for ball retrieval? One that coaches might want to use?

I skype my mate Tony, who has a lot of RC stuff and a lot of experience with software as well.

“Forget video,” he says, “You want to use RFID.”

“I’ve been trying it out on the cars, and it’s pretty good.”

“Radio-frequency identification (RFID) uses electromagnetic fields to automatically identify and track tags attached to objects.” – Wikipedia. Today, the market for RFID is around USD10 billion annually and growing. It’s the technology in that device the toll company expects you to attach to the windscreen of your car, for example. It’s in clothes – clothing companies are in trouble for tracking the movement of customers by tracking the movements of tags in the clothes they buy. It’s in event tickets, in every kind of proximity card. All over the place. It’s often a transmitter about the size of a grain of rice, worth between 5c and 15c a unit.

some of Tony's cars
I have a few concerns about RFID I’d like to discuss with Tony, but it’s too late. I’ve fired up his enthusiasm and he’s gone, out to the shed to get a car and some RFID tags and have a play with it all in the backyard at home.

So what are my concerns? Pretty obvious ones – how’s it going to work, and how much will it cost? And has someone already done this?

Let’s start with the last question. Time to engage the inter-bubble.
Wallace E. Bolden Jr.
In 2012, Wallace Bolden applied for a US patent for a system that would track balls using RFID. His application includes some pictures of how the system works. I am sure though, that he’s not actually built a working version of this system. Don’t you have to do that if you’re making a patent application? “You can’t patent an idea”, right? So I look him up in the bubble, so I can ask him directly, and guess what! A friend of mine is a contact of a contact of his on LinkedIn. We’re so close. So I decide to leverage our LinkedIn (three degrees of separation) connection – I email my friend and ask him to introduce me to his contact so his contact can introduce me to Wallace so I can ask him if he’s actually built a working device or if it’s some kind of ambit claim ripe for reverse engineering once someone else builds a working system and he wants to sue them for patent infringement. In a nice way though.

Two weeks later, and my friend (well I thought he was my friend) still hasn’t responded. Perhaps he’s not as enthusiastic about the power of (yet another social network, this one called) LinkedIn as I am.

But that’s okay, because in the meantime, I’ve broadened my search, and come up with a load of sites citing various systems using RFID to locate, in particular, golf balls. Ominously, ‘Radar Golf Inc.’, the one time lead innovator in the field, seems to not exist anymore.

Following these leads through I eventually fix on something solid. At the ‘Fraunhofer Institute’ in Germany, they’ve got RFID working in numerous sporting applications. A collaboration with Adidas has produced systems actually being used by a German soccer team. RFID tags on balls and monitors placed around goals can determine with never-before-ascertainable accuracy whether or not the ball has actually crossed the goal line. And other applications tied more directly to team performance. Other systems are being trialed in rugby and ice hockey, and discussions are under way with skiing, basketball and golf entities.

The guys at the institute who are heading up this work, their email addresses and phone numbers are right there on the net, so I bang off an email, without even cross-referencing LinkedIn.

And get no reply.

From the German guys that is.

On the other hand, I’m hearing a lot from Tony, who thinks the idea is a goer, and is urging me to assert some kind of intellectual property right over the whole concept (perhaps I’ll ask Bolden if he has a friend like Tony if and when we ever actually catch up).

Tony’s telling me the system he’s thinking of is going to work just fine. I ask him how he thinks the tags will attach to tennis balls. He’s assumed I’ve got that bit sorted out. I point out that if we want to use RFID with the balls, we’re going to have to get the tags inside the balls somehow. He agrees we have to do that. I also express concerns about the rapid acceleration a tennis ball undergoes, being thumped really hard with a bat designed for thumping it really hard.

Tony confuses acceleration with velocity, and tells me that the maximum velocity of a tennis ball is about 240km/h. I’m trying to explain that going from high speed in one direction, a lot of force is applied to a tennis ball to get it to accelerate extremely rapidly in the opposite direction. But it’s too late. Tony’s gone, back to the test site (his backyard and shed), where he’s assessing how accurately he can ascertain the location of an RFID tag on an RC car traveling at more than 200km/h.

There’s an audible doppler effect in the tone of the little car’s motor, as it shoots past the phone, and I duck involuntarily.

A little while later, Tony wanders back over to the phone, and says he thinks there might be something in what I was saying about acceleration. Says he’s had a few red wines this evening.

“Oh by the way,” he says, “You know I was at that concert too, the one you wrote about a few weeks ago?”

“And Sue,” he says, referring to his wife, “Was too, though I didn’t know her then. But we bought tickets.”

A few people have been telling me they were at the concert too, all of them stressing that they bought and paid for their tickets. Only one of them though able to recall more detail than me about the actual performance.


I’ve been thinking a little more about Tony’s proposal that we use RFID, and I can think of several different tennis scenarios where the technology might be applied:

1.       At the professional level: a working RFID system could outperform the ‘Hawk-Eye’ system – could in fact replace the human line call system entirely. Hawk-eye has limitations – speed, accuracy, infrastructure and price. An RFID system would be less defined by these limitations.

2.       At the coach level: and this is where I originally envisaged the technology being applied, a ‘robot ball boy’ that works with a coach while he’s giving a lesson, to collect and return balls.

3.       At the personal level: for about $80 you can buy a system to help you locate your possessions using RFID. You attach a tag to your item, and when you need to, use a ‘direction finder’ to point you towards it. A similar system could track down those balls the kids hit over the fence, lose in the bushes, etc.

The ITF (International Tennis Federation) website clearly defines the properties of a tennis ball, and nothing suggests that an intra-ball infrastructure supporting components of an RFID location tag system contravenes any of their criteria.

It all seems reasonable. Why haven’t the Germans emailed me to say they think so too? I decide to call them and find out – perhaps my emails all ended up in the junk mail.

The first two days I call, no-one answers. But on the third day:

René Dünkler
I’m talking to René Dünkler at the institute. We talk for about twenty minutes, during which it becomes increasingly clear to me that he has actually received and read my email, and ignored it. Setting this aside, we have a good discussion about the application of the technology to tennis. Dünkler is clearly proud of the work they’ve done in other sports, and the collaboration with Adidas. 

Dünkler describes my thinking about RFID and tennis as an ‘interesting project’ that the institute would like to take on. He thinks it would be possible but problematic to use RFID to locate tennis balls. The problems link not to the ability of the technology to quickly and accurately locate balls, but to attaching the tag (or tags) to the ball. Tennis balls are empty inside, and undergo rapid acceleration. 

We agree it’s a line of thought worth pursuing but one that would require the participation of a tennis ball manufacturer, not to mention substantial funding. 

But think for a minute: tennis balls is big business, and all the manufacturers make a very similar product. Incorporating RFID technology into tennis balls would substantially differentiate one manufacturer’s product from the others, at least until the others catch up. 

an Adidas tennis ball
Which manufacturer? Back to the bubble I go. While looking for contact information for manufacturers, I find that Adidas released a ‘tennis ball’ in August 2016, together with pictures of tennis balls with Adidas logos. Ultimately I realise the Adidas ‘tennis ball’ is actually the name of a green shoe cross-promoted by an American rapper, and that none of the Adidas-logoed balls are actually an Adidas product. 

Nevertheless, Adidas is a sports manufacturer with a technological bent, they make balls, and they have established links with the Embedded Systems folk at Fraunhofer. It’s worth a look. 

not an Adidas tennis ball
The inter-bubble throws up Jon Werner – Innovation Explorer at Adidas in the USA – as a good person to talk to. So how do I track him down? The corporate website is clearly going to be no help. I find him on LinkedIn, where he’s left ‘contact info’. One of the contacts is the Adidas website, but the other suggests Twitter. His Twitter profile describes him as ‘friend to all (unless you are evil)’. So I message him, and get a message back that says since he doesn’t ‘follow’ me on Twitter, I can’t message him. So I tweet, a few tweets because it doesn’t fit in one, roughly: ‘Hi friendly Jon. Wanted to chat about RFID and tennis balls, but can’t message you, so I’m sending these tweets. Please drop me a line when you have a minute. Surely there’s a better way to communicate than this?’ I stare at the screen waiting for the immediacy of (yet another social network, this one called) Twitter to take effect. For about a minute I do that, then head off on a different tangent when nothing happens. 

I'd never actually bothered to ask tennis coaches if they’d like a robotic ball boy. I call one coach I know, and email another. Matt McDonald, the professional at Red Hill Tennis Club wants one, wants to order one now. I explain to him that I need to get a tennis ball manufacturer on board. He gives me his contacts at Babolat and urges me to follow up. The other coach, Mark Fabian, is not a full time professional, more a roving consultant. He says he wouldn’t buy a robot, as he’s not based anywhere he could keep it..But he's sure that clubs would definitely want them. He starts giving me reasons why coaches would love them – allows coaches to spend more time coaching, keeps parents and other ball collectors off the court (but especially parents), could be integrated with a tennis ball machine to keep the balls coming, etc. 

Okay, so there’s a market for the coaching/club level product, if it costs no more than a few thousand dollars and works properly and reliably. That’s important. Now I’ve got to get a manufacturer on board, one prepared to work with Fraunhofer to turn my idea into reality. This idea has two parts – one is building the ball-with-RFID, the other is building the systems to track it, to varying degrees of sophistication – personal, club and professional, with club level the main focus. 

Tony’s back. He thinks we should go it alone. He starts telling about the various factories in India making tennis balls for international brand names. (I’m in India, Tony’s in Australia.) 

Tony is listing all the Indian cities with tennis ball factories when my desktop tells me Jon Werner sent me a personal message via Twitter 22m ago. Sorry Tony, gotta go, I’m 22m behind the green ball.

Jon Werner
Werner and I spend 15m sending PMs back and forth via Twitter. To cut to the chase, Werner sees this idea as something that’s going to come of age simultaneous with an RFID internet of things that will itself explode on the world once mobile phones include an RFID tracker. Werner has locked onto my third, personal, application for RFID in tennis balls – helping individuals locate their tennis balls, at the court, or even in the shed, house, or car, depending how errant or disorganised that individual is. He thinks this day is precipitant now that Qualcomm (a phone components manufacturer) has purchased NXT (a maker of other types of electronic components) in a deal worth $47 billion. But he doesn’t think it’s something Adidas is very interested in. However, he also revs me up, telling me if I have an idea, I should ‘go for it’. The little I know of entrepreneurialism suggests that having a good team is critical to success. 

Tony and I are not a good team. Tony’s now giving me instructions derived from GoogleMaps on the most sensible route to the tennis ball factory closest to where I live in Delhi, two and a half hours away in a city called Meerut. The instructions, nuanced to account for varying traffic conditions, suggest I leave home at 4.30am. 

“Oh by the way,” he says, “I’m sending you a surprise.” 

I can’t wait. 

But I’ll have to, and use the interim to call Matt McDonald’s mate Alex at Babolat. They’re a casual lot at Babolat – don’t even use surnames. I talk to Alex for a while. He’s a bit mystified by the whole story and my motivation, a little bit sceptical. But he asks me to put the detail in an email and promises he’ll move it on to the relevant product development person at Babolat, and that I should hear back in the not-too-distant future. 

And that’s about the end of this odyssey, if it's ever going to be posted. 

Except for one loose end:

Tony’s back. 

“Did you see it?” he asks. 

Did I see what?

"The RC car on your front verandah." 

There is, in fact, an RC car on the front verandah. 

“Do you mean the one with ‘RFID Tennis ball project experimental vehicle #1’ stencilled on it?” 

“Are there two RC cars on your verandah?” he asks, rhetorically, as I have to concede there are not. 

Tony explains that the little vehicle navigated its entire journey from Canberra to Delhi, in less than a blog’s breath, relying solely on RFID signals. Which I find hard to believe. 

"Now,” he says, “Are you ready for your surprise?” 

The car on my verandah isn’t surprise enough?

Suddenly, I’m hit in the forehead with a tennis ball. I recover, just in time to duck as the car fires a second furry green cannonball at me, and then a third, and … 

“TONY !!!!”


No comments:

Post a Comment