GoPro partners with WTC, and developed a custom version of the Hero5. This custom version has no buttons. It has an attached supplemental battery to give it 8 hours of recording time. It’s attached to the seatpost of every bike on bike check-in by a race official. It has a rearward FoV. Everyone takes the same aero hit (more or less).
Using the Smart Remote control interface already developed by GoPro, each Hero5 is activated automatically upon leaving T1. Upon entering T2, an automatic upload is triggered over an encrypted link. 15 minutes later (or whatever) the GPS record is run through a Strava Flyby-like algorithm that flags potential drafting vs every other GPS file that you ever come within a certain distance of - say 20 meters. Because GPS isn’t perfect and there may been circumstances under which an official would let it go (e.g. some safety situation), the algorithm then cues up a 45-second clip from the GoPro of the guy ahead of you, and sends it to a team of officials sitting at workstations. If the officials decide it is, indeed, drafting, they flag it.
Enforcement takes place during the run. At the halfway point of the run, there’s a Hertz #1 Club Gold-like sign with your name and #. If you’re on it, you enter the penalty box or get DQed.
But wait, there’s more. What do GoPro and WTC get out of it (other than marketing for GoPro and a cleaner race).
There’s all that video, owned by WTC. After the event they go back to the Strava Flyby algorithm and compile a video of just you every time there’s a good view of you, e.g. when you’re getting passed or are passing. They offer that video to you for, say $100 or something. So if you killed the bike leg like every 'twitcher should you can have continuous loop of every pass on your computer as a screensaver. So you can just stare at it for hours while getting sexually aroused watching your bike prowess and slammed position. Nevermind that you walk-jogged for the next 4 hours.
And for the pros it gives them excellent footage (maybe even realtime) of the race in progress. You could offer a Web service. I want to see Frodo’s footage live, with overlayed speed and power. That’d be cool.
Capitalism. Technology. Marketing. What’s not to like?
Never happen, unfortunately. Until Star Trek is more reality than fantasy.
All kidding aside, have you done the math? Assume 1,000 racers. Assume 100 passes per racer. That’s 100,000 passes that will trigger a review. That’s 75,000 seconds of video to be queued up and watched. Like 20 hours of video, not even counting the time to find it and queue it up.
Then there’s the upload and storage requirements. What kind of infrastructure would it take to stream that much video from the bike to the server(s)? How long would it take? How much disk space does 5,000 hours of high-quality video consume?
Then do the math for 2,000 IM-distance racers.
You could work with Strava to enhance their fly-by software algorithms to filter out all intersections that lasted less than, say, 30 seconds in order to reduce the reviews. I would imagine that some Big Data analyses would be pretty interesting, in terms of rider density and course position, etc.
Never happen, unfortunately. Until Star Trek is more reality than fantasy.
All kidding aside, have you done the math? Assume 1,000 racers. Assume 100 passes per racer. That’s 100,000 passes that will trigger a review. That’s 75,000 seconds of video to be queued up and watched. Like 20 hours of video, not even counting the time to find it and queue it up.
The Cloud, man, The Cloud.
Only half joking. Strava is probably doing thousands of Flyby calculation at any given moment. 10’s of thousands during peak workout hours. I’m sure they’d license their scaling solution. (GoPro drafting enforcement p/b Strava).
Similarly with video. Youtube handles millions of location-seek (where you drag the bar to some part of the video) requests to their “trending” videos per second. 1000 racers should be something like pocket change.
Disc space is easy. You can get 8 hours of 1080p on a 64GB card. Times 1000 = 64 Terabytes. A 6TB drive is about $400. So 11 of those. Call it 22 for a safer RAID configuration. Add in a NAS box and it’s about $10K all-in. Pocket change.
Just for illustration of the problem, Trimes.org has published a video and some photos where is spite of any argument of angle not being great etc, drafting is obvious. Paper in French, but the photos at the end of the page don’t need translation http://www.trimes.org/...va-bien-tout-va-mal/
Egads! That looks more like a mass start race. I always wondering why triathletes did group rides on tri bikes, since drafting is illegal in most races. Kind of defeats the training purpose, though from the photo it now makes sense…
I like the video idea, but another option that can be easily implemented today is to have multiple people with cameras randomly along the course to record the riders. It wouldn’t catch all drafting, but would probably be more effective than just having moto refs on the course.
I like your idea in general, but given the inaccuracies of GPS (easily 10m or so), I would think you’d have a lot of false positives or misses.
That’s why there’s video.
Also, GPS accuracy is generally reported in term of absolute accuracy. But what matters here is relative accuracy, e.g. the distance between two GPS receivers. That might sounds worse at first. Because if both are 10m off in opposite directions then you have a 20m error in your distance measurement.
But, fortunately, some types of GPS error will tend to affect both receivers in the same way. E.g. they might have a 10m error vs. an absolute position, but it’s like to be the same bias - say 10m to the left. So a distance measurement might be quite accurate - much less than 10m Possibly less than a meter most of the time.
However other types of GPS error will affect each receiver independently, so certainly some testing would have to be done to understand how accurate it really is.
I like your idea in general, but given the inaccuracies of GPS (easily 10m or so), I would think you’d have a lot of false positives or misses.
That’s why there’s video.
Also, GPS accuracy is generally reported in term of absolute accuracy. But what matters here is relative accuracy, e.g. the distance between two GPS receivers. That might sounds worse at first. Because if both are 10m off in opposite directions then you have a 20m error in your distance measurement.
But, fortunately, some types of GPS error will tend to affect both receivers in the same way. E.g. they might have a 10m error vs. an absolute position, but it’s like to be the same bias - say 10m to the left. So a distance measurement might be quite accurate - much less than 10m Possibly less than a meter most of the time.
However other types of GPS error will affect each receiver independently, so certainly some testing would have to be done to understand how accurate it really is.
yeah, i don’t disagree, but it also assumes both receivers are tracking the same satellites. In places and times where there are lots of sats in view, you might have receivers with slightly different sets of satellites. Here in Ohio it’s not uncommon to have a bunch of satellites in view most of the time. Which satellites had the best angles for the receiver might also change over the course of a couple hours for everyone to get on their bikes.
Which satellites had the best angles for the receiver might also change over the course of a couple hours for everyone to get on their bikes.
That’s true. But for the 20-25 seconds of time necessary to conclude drafting might have occured, the constellation is quite likely to be the same for two receivers <20m apart.
It’s true that accuracy could change dramatically with position on the course and time of day. But since GPS reports its own accuracy reasonably well, it should be possible to just ignore location or times with really bad data. Just ignore the data if both recievers don’t see 5+ satellites or the “dilution of precision” numbers exceed some #.
Which satellites had the best angles for the receiver might also change over the course of a couple hours for everyone to get on their bikes.
That’s true. But for the 20-25 seconds of time necessary to conclude drafting might have occured, the constellation is quite likely to be the same for two receivers <20m apart.
It’s true that accuracy could change dramatically with position on the course and time of day. But since GPS reports its own accuracy reasonably well, it should be possible to just ignore location or times with really bad data. Just ignore the data if both recievers don’t see 5+ satellites or the “dilution of precision” numbers exceed some #.
It would be an interesting experiment, anyway. Would have to have 2 of the same device with all the same firmware and the same GPS chipset to see how close they match.
Why GoPro? They stopped their sponsorship of Kona.
Put it on Garmin. Their cameras have GPS, but you make the bundle include their varia radar. That way, anytime the varia senses someone is in the box, it’ll start recording. Less battery, less total footage, less time to screen since it is more accurate than GPS.
Normal varias are calibrated to ignore cyclists, but if a bike is going fast enough up on you, it’ll trigger the alert. So, they could turn down the discrimination and get all the bikes (and drivers on the course…which would be good info to have as well).
The course could have multiple painted boxes on the road so that you could ‘calibrate’ you draft zone in the FOV…in case the camera is pitched up or down.
Plus, with a couple thousand radars pinging around, any streetracers with radar detectors will stay off the course.
Why GoPro? They stopped their sponsorship of Kona.
No reason, just partial to them since the CEO was a classmate of mine.
Could just as easily be Garmin. Decent idea with the radar, but everything gets more expensive with radar. GPS and cameras are dirt cheap relatively speaking.
Instead of giving us passive RFID tracker timechips, everyone gets a timechip with a GPS device with real time tracking via LTE connectivity. That tech could be possible very soon. Samsung just released a wristwatch with LTE. If that’s possible, so is this. Biggest issue would be battery life.
Think of it as your garmin, but with just time and location data, no other sensors, no display, and it sends data in real time rather than record it locally. Even with 1s recording that’s a tiny amount of data totaling a 5-10 MB per athlete in an Ironman. Because it is just text.
The 10m GPS accuracy won’t matter when the data is uploaded and analyzed. With a datapoint every 1s it would be trivial for a computer to crunch those numbers and find the pelotons. This should be automated because it’s precisely something computers are good at and people are bad at.
On the referee side, they could have a device, lets say a tablet, with a list of drafters to tag and send to penalty tents.
Obviously like anything else new, this should be tested before relying on it for timing, and tested some more before relying on it for finding drafting.
Bonus, your friends and family could watch your progress in Google Earth or similar.
I think your numbers need some common sense applied to them.
I don’t believe it’s remotely realistic for 1000 people all to pass 100 people. Sure, there’s one guy that passes 100 people, but half the people never pass anyone. Half the people do the passing, and the other half are getting passed. That means that notionally, your 1000 people gets dropped to 500. That halves the problem before we even get started.
Look at me for example. My last 2 IM races I finished almost exactly the 50th percentile. I swam a tad over 60 minutes both times. That gets me out of the water about top 20%. In a race of 2000 people, I get passed about 200-250 times in just over 6 hours. (yes it’s depressing, yes you monster cyclists need to learn how to swim). If I pass more than 3 people on the bike (that aren’t stopped on the side of the road for various reasons), I’m having a good day.
The whole checking passing manoeuvres is quite small. To get passed on the bike, the person being passed has to be a good swimmer and crap cyclist. Most triathletes are relatively balanced across the 3 disciplines. Things tend to stretch out over the day. Good swimmers tend to be good cyclists and runners too. Sure there is movement through the pack, but not much. Mostly, things stretch out.
There are relatively few people like me, that swim OK ride like crap and run even worse. I’d suggest that the 1000 riders passing 100 people, for 100,000 events, is more likely to be less than 5000 events. Rather than either of us guessing, it’s a relatively straightforward exercise to check how many passing events took place in a given race. Overall race rank coming out of the water to T2 rankings. The improvement by T2 is passing events that could be checked. If the problem is 100,000 events, then that’s a much bigger issue to solve than if it’s 5000 events. 5000 would be easy to police, 100,000 events, not so much.
Which satellites had the best angles for the receiver might also change over the course of a couple hours for everyone to get on their bikes.
That’s true. But for the 20-25 seconds of time necessary to conclude drafting might have occured, the constellation is quite likely to be the same for two receivers <20m apart.
It’s true that accuracy could change dramatically with position on the course and time of day. But since GPS reports its own accuracy reasonably well, it should be possible to just ignore location or times with really bad data. Just ignore the data if both recievers don’t see 5+ satellites or the “dilution of precision” numbers exceed some #.
GPS plus Glonass plus a large enough receiver antenna (think Garmin Edge 1000 sized) plus carrier phase measurement and you can get into the centimeter range for accuracy.
It’s not off the shelf but all the existing technology is out there. I think it’s very doable from a tech standpoint. I could also see the market argument. I’d gladly pay $100 for pics plus video of a big race. If Ironman purchased such a system and used it 30x per year, 2,000 races per year, $100 per competitor… that’s $6,000,000/year in additional potential revenue. The capture rate would be way below 100% but I’d bet it’s still high enough to pay for itself in the first year.
You could even automate the draft distance check. Have a distinct shape on the helmet number (or even better multiple) and enforce number placement on the helmet. Since every gopro will have the same lens, have a check where is the number is taking up greater than a certain number of pixels for a certain amount of frames, you have drafting. Or maybe the sticker with the shape goes on the head tube.
With the advent of Quark’s new data recorder with phone output, why worry about the issues of video/sync/etc and use something like Garmin’s Varia. I don’t know if Varia can transmit distance warning over ANT, but if it can be set up that way just have it trigger an alert if the rider is too close and beam it back to headquarters in real time. No need to know how close, if you cross the warning range and there’s another rider ahead you get docked every time it crosses. From the perspective of too many riders, IMO it only really makes sense for the top 500 riders, and people know who the top 500 are going to be. There’s no reason MOP/BOP need them IMO - while it’s inconvenient, the meat of the problem manifests itself in the front more than the rear.
Or go with your bazillion dollar technical solution…
Well I became a roadie. Problem solved. Just trying to help out the triath-a-letes. Because, damn, those drafting packs.
But WTC is about extracting as much money as possible from affluent middle-aged age-groupers. Hard to do with small fields and staggered wave starts strain against permit time limits. And I think my solution pays for itself in marketing and revenue generation. A GoPro 5 costs around $400, but I think they have a healthy margin - probably around $100 to manufacture. So around $100K in GoPros for a 1000 person race. But those are re-usable. Factoring in development labor and other hardware and software licensing, I’d estimate around $3M to develop a package that can be moved from IM to IM. Or roughly the the revenue to Active from their “processing fee” for a single IM.
Just trying to help out the triath-a-letes. Because, damn, those drafting packs.