Author Topic: Apple iWatch with Dawn Patrol SURF TRACKING  (Read 7252 times)

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1920
  • downwind dilettante
    • View Profile
    • Email
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #60 on: January 26, 2021, 03:42:48 PM »
Garbage in, garbage out.  Telemetry scrolled real time (like a scope on a hospital heart rate monitor) will show you the error of your ways.  Flatlines at 18Hz is one thing, at "~1Hz" would give new meaning to the concept of  "data smoothing." 

Here is an example of a real time speed graph, overlayed on video.  It scrolls thru 10 seconds of telemetry (synched automatically on the video file, via gopro).  Your gopro Max will do a speedometer overlay with the same data.  (That's what Robby Naish used on his wingfoil speed at Maliko video).  Both cams sample at 18Hz.

See all the missing points on the line?  That's 180 points on that line, IF none are missing. Flatlines are missing points.  It's very common among GPS telemetry.  (That's why the DJI pros go up to 100Hz).   

Your iWatch is struggling to send 10 points to the Waterspeed app.  Then the app averages it all.     

The iWatch antenna is mounted to a moving object (wristwatch), Gopro antenna is mounted to my head.  We have excellent GPS signals here, as do you.  The helicopter guys have a nifty site to prove it.   https://www.uavforecast.com/

So, those air drops Haley is doing may very well be 34mph.  But the data points you are using are weak at that rate of change.  Therefore, top speeds spewed by the Waterspeed app are wildly speculative.  You know it better than I do with your math background.

When you put all the data alongside the video, it's easy to see.       
« Last Edit: January 26, 2021, 03:48:17 PM by LaPerouseBay »
Support your local shaper

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1920
  • downwind dilettante
    • View Profile
    • Email
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #61 on: January 26, 2021, 04:17:15 PM »
Beasho,

Here is the full video.  In my sport, we aim to flatten that speed line.  We avoid dropping straight down the waves, unless we can "make the next section" (to use surfing parlance).  That's why I whipped up that graph in Racerender.  If the speed graph drops precipitously, I'm about to do something wrong....   

In surfing, a scrolling speed graph will help illustrate the accelerations as you turn down the line.

That said, it's important to realize that all those fancy numbers are relative to the ground, not the surface of the water...  The effect of angling across a wave would be easy to see.  You should do it on your foil at Mavs.  I'll send you the template for that graph, it's easy to adjust for your speeds. 

That long ride Robby did averaged 18mph.  He was flying across the wave, just as Laird does on foil at Nazare'.  Nothing against air dropping, then straightlining big Mavs, it's all amazing to everyone. 

If you really want to claim speed across the water (or air) using GPS - accurately, you gotta go mechanical.  Rowers have been using them forever.   

https://youtu.be/B_FlTgM_Xw4
Support your local shaper

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1920
  • downwind dilettante
    • View Profile
    • Email
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #62 on: January 27, 2021, 01:51:05 PM »
Here is a link to the impeller, with a few comments about the limitations of GPS. 
https://www.rowinginmotion.com/designing-measurement-system-rowing-boat-speed/

Here's Robby on a few videos I referenced.

Average speed is over 18mph.   
https://youtu.be/M4LQqMhdg5E

Gopro on his head, with the speed overlay (GPS at 18Hz).  Gopro cams have the gauge in their software somewhere.  It will be far more accurate than an iWatch and Waterspeed app, recording "~1Hz"  Particularly for air drops.  Spikes like an air drop are when I get very skeptical.  Hence the running graph.  I've seen some weird stuff on my 1Hz running graphs.   
https://youtu.be/0fqaT7LOCvY?t=193
Support your local shaper

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #63 on: January 28, 2021, 07:50:29 AM »
Garbage in, garbage out.  Telemetry scrolled real time (like a scope on a hospital heart rate monitor) will show you the error of your ways.  Flatlines at 18Hz is one thing, at "~1Hz" would give new meaning to the concept of  "data smoothing." 

Here is an example of a real time speed graph, overlayed on video.  It scrolls thru 10 seconds of telemetry (synched automatically on the video file, via gopro).  Your gopro Max will do a speedometer overlay with the same data.  (That's what Robby Naish used on his wingfoil speed at Maliko video).  Both cams sample at 18Hz.

See all the missing points on the line?  That's 180 points on that line, IF none are missing. Flatlines are missing points.  It's very common among GPS telemetry.  (That's why the DJI pros go up to 100Hz).

Ahhhh now I get it you are demostrating crappy technology.  Here is an example of Data extracted from TRACE.  It was my standard for 5 years. 

TRACE NO FLATLINES EVER!  I know because I downloaded the data.  30,000 records per session 

Another example of a standard wave (2nd view).  The Blue Charts show the MAXIMUM and MINIMUM for every SECOND.  A Flatline would have dropped out of the the view. 

Having recorded and downloaded 5 million data points the TRACE never glitched. 

« Last Edit: January 28, 2021, 07:56:16 AM by Beasho »

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #64 on: January 28, 2021, 07:58:45 AM »
Here was an article I wrote for Microsoft's BI Semantic model, the technology and the code used to extract the data.  There is an interactive wave selection showing these graphics for every wave.

No Flat Lines.

https://powerpivotpro.com/2018/02/power-bi-used-for-surfing-fun/
« Last Edit: January 28, 2021, 08:40:41 AM by Beasho »

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #65 on: January 28, 2021, 08:40:01 AM »
When I wrote that article I offered to share my "APP" with anyone that wanted it.

Take It!  Process your data.  Tell us what you find.  Share the pictures . . . .

NOTHING!!!!! Crickets.  Not a single taker anywhere in the world. Not on the P3 website, Not here on the StandupZone.  Not on https://www.seabreeze.com.au/forums/Stand-Up-Paddle/Foiling NOTHING.

I recently read an interview with Steve Jobs from 35 years ago.  And he succinctly explained why.  And why the Apple watch is great. 

It gives you instant feedback on your wrist "How long was that wave?"  "How fast did I go?"  No code, just pretty pictures of your rides that you can share with your friends.  Instantly. 

http://reprints.longform.org/playboy-interview-steve-jobs

"Some people are saying that we ought to put an IBM PC on every desk in America to improve productivity. It won’t work. The special incantations you have to learn this time are “slash q-zs” and things like that. The manual for WordStar, the most popular word-processing program, is 400 pages thick. To write a novel, you have to read a novel—one that reads like a mystery to most people. They’re not going to learn slash q-z any more than they’re going to learn Morse code. That is what Macintosh is all about. It’s the first “telephone” of our industry. And, besides that, the neatest thing about it, to me, is that the Macintosh lets you sing the way the telephone did. You don’t simply communicate words, you have special print styles and the ability to draw and add pictures to express yourself." - STEVE JOBS
« Last Edit: January 28, 2021, 09:08:03 AM by Beasho »

PonoBill

  • Cortez Bank Status
  • *****
  • Posts: 24292
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #66 on: January 28, 2021, 09:06:25 AM »
Yes, if you want to record water speed you need some way to measure that directly. When I was doing testing for the KeNalu paddles I used an RC modeling data recorder and one input was a free-spinning impeller with a chopper wheel that triggered a hall effect sensor. Another input was a GPS module that included direct speed measurement via doppler. At one time you could buy handheld GPS devices and GPS modules that included doppler speed recording, but I haven't seen one for years. The general market doesn't care that much about speed accuracy. The GPS speedometer I installed in my Moho project does speed calculation from location--that from a device whose only purpose is speed measurement.

Over about three months of intensive testing, and the occasional sporadic testing later, the impeller speed and the GPS speed never agreed, even with doppler measurement. Even testing in flatwater the speeds never agreed, in the river the difference was huge, like +/- 5mph.

Location measurement via GPS without doppler is inherently inaccurate, both because of DOP issues (HDOP, PDOP, GDOP) and atmospheric issues. Fortunately, the feds turned off the time signal randomizers in 2000, so it's no longer wildly inaccurate, but it's still about +/- 5 meters. Increased sampling can help, but that doesn't change the simple fact that every measurement can be off by as much as +/-5 meters, and the distribution is somewhat random. You can be standing still and the GPS thinks you just jumped 10 meters in some fraction of a second. Raw GNSS data can be manipulated to give an accuracy of about +/- 1 meter. I don't know if GoPros, iWatches, Trace or any of the stuff you guys are using does that, but that's still a fairly shitty ruler.
Foote 10'4X34", SIC 17.5 V1 hollow and an EPS one in Hood River. Foote 9'0" x 31", L41 8'8", 18' Speedboard, etc. etc.

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #67 on: January 28, 2021, 09:22:39 AM »
Raw GNSS data can be manipulated to give an accuracy of about +/- 1 meter. I don't know if GoPros, iWatches, Trace or any of the stuff you guys are using does that, but that's still a fairly shitty ruler.

One way to test the LATITUDE - LONGITUDE accuracy - ROCKS!!!!


See these tracks.  I have highlighted in RED where the Rocks are: 

D = Dent Rock (I landed on this one time)
C = Crown Rock
M = Mushroom Rock

These tracks were exactly as I rode them through the rocks.  Within that 1 meter specification. 

When data is extracted you can get the LAT, LONG and exact time (I assume the watch is pretty precise with time).   Speed can be reverse engineered using the HAVERSINE equation.

This once again points to the accuracy of these devices and something that can be verified by HARD OBJECTS. 
« Last Edit: January 28, 2021, 09:25:49 AM by Beasho »

Tom

  • Cortez Bank Status
  • *****
  • Posts: 2926
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #68 on: January 28, 2021, 10:19:02 AM »
How about four surfers riding a wave while windsurfing for 1.65 miles? This was in San Carlos Baja during a major Northwest swell in the spring sometime around 1990. My friend Greg Miller and three others took off on a couple of waves in the same set at Bombora and rode them all the way to the fish camp. Somewhere I've got the windsurf magazine which documented the ride, but I can't find it easily. I used Google maps to estimate the distance. I can't prove it, but I've talk to most of the surfers.

PonoBill

  • Cortez Bank Status
  • *****
  • Posts: 24292
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #69 on: January 28, 2021, 10:57:34 AM »
Interesting that you'd mention Haversine, I'm using that equation in my TowBot code. Or at least in the current iteration. I think there's an easier way, it looks like the Meshtastic software is making a very simple calculation for distance and heading between two GPS equipped modules talking to each other via LoRa. I don't understand it yet, but I'll figure it out.

One meter accuracy is fine for saying where something fixed is located, but for speed calculation where the speed is low, it can be a problem. Averaging a truly random set of coordinates with a known 1-meter accuracy improves the practical accuracy of the location even though the resulting accuracy can't really be determined. A truly random error could yield +1 meter to the left for 1000 measurements. Extremely unlikely, but possible. For a speed measurement, the number can jump around in the +/-1 range for every measurement. Assume the worst case where it alternates the jumps +1 and -1 for every measurement set and the actual speed is 10mph, about 4.5 meters per second. At a sample rate of 10hz that's .45 m/.1 sec. Two alternations around 0 at a sample rate of 10hz yields -.55m/.1sec and the next 1.55m/.1sec. or -5.5m/sec for the first and 15.5m/sec for the other. I'm doing the math as I type so I probably fucked up somewhere, but the principle is reasonably correct. Of course in practical terms that doesn't happen, but this is the reason that GPS calculated speed is not used for speed records.

There's a good paper on this here, which will probably be read about as often as yours. Meaning -not often. they said there'd be no math.
https://www.e-education.psu.edu/geog862/node/1786
Foote 10'4X34", SIC 17.5 V1 hollow and an EPS one in Hood River. Foote 9'0" x 31", L41 8'8", 18' Speedboard, etc. etc.

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1920
  • downwind dilettante
    • View Profile
    • Email
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #70 on: January 28, 2021, 01:02:27 PM »

Ahhhh now I get it you are demostrating crappy technology.  Here is an example of Data extracted from TRACE.  It was my standard for 5 years. 

TRACE NO FLATLINES EVER!  I know because I downloaded the data.  30,000 records per session 

Another example of a standard wave (2nd view).  The Blue Charts show the MAXIMUM and MINIMUM for every SECOND.  A Flatline would have dropped out of the the view. 

Having recorded and downloaded 5 million data points the TRACE never glitched.

That's hilarious.  Now I know you are messing with me. 

Here is a graph using a program even older than either your iWatch/Trace, (at 5Hz) and the iWatch/Waterspeed (at "~1Hz.")  This is Garmin's old "Training Center."  I prefer it to the newer Garmin graphing, which is dumbed down for the social media market.  All the new consumer grade stuff is.  They gotta follow the market.   

From that video I linked above: I'm zooming along, then stop to wait to meet up with a fellow paddler.  I'm just sitting there, floating along.  At 18Hz.   See all the dropped signals?  Smoothed even more in the second pic...  If you want to put your faith in that little iWatch at 1Hz, air dropping, have at it.  I'm not buying all those bar graphs.  That data is smoothed.  That's a whole lot of 1's and 0's in your little bars...

GPS on the water is pretty rough to begin with.  1Hz is weak sauce, even if you hi tech watch isn't dropping the occasional signal... 5 million points and you never had a glitch?  Dig deeper. 
Support your local shaper

PonoBill

  • Cortez Bank Status
  • *****
  • Posts: 24292
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #71 on: January 28, 2021, 03:47:38 PM »
Oops, wrong link. Nice reference to general GPS stuff though. Here's the right one.
http://bioresonant.com/dl/dl.htm?name=HighAccuracySpeed.pdf

Yeah, the link is weird but it should work. It just downloads the .pdf. AFAIK it won't install any vicious malware.
« Last Edit: January 28, 2021, 03:49:19 PM by PonoBill »
Foote 10'4X34", SIC 17.5 V1 hollow and an EPS one in Hood River. Foote 9'0" x 31", L41 8'8", 18' Speedboard, etc. etc.

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1920
  • downwind dilettante
    • View Profile
    • Email
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #72 on: January 28, 2021, 07:37:29 PM »
Another example of a standard wave (2nd view).  The Blue Charts show the MAXIMUM and MINIMUM for every SECOND.  A Flatline would have dropped out of the the view. 

Here is some more bad news about those data points from an iWatch.  From this page.  https://www.letsrun.com/forum/flat_read.php?thread=9954519

----------------------------------------------------------------------------------------------------------------------------
I just bought an Apple Watch 5 and returned it. Here's what I found:

- Most watches on the market use 1-second GPS fix intervals

- The Apple Watch uses an interval that "varies." The definition of that is proprietary, but from what I've seen online (and anecdotally validated through using it) is that the the Apple Watch is taking a fix roughly every ten seconds then using an accelerometer to interpolate position. Apple has a high-res screen (big battery drain), so the reduced GPS fixes conserve battery life.

- The Apple Watch uses a generous GPS smoothing function. This makes really pretty lines on a map, but it isn't very responsive to pace changes, especially when combined with the increased GPS fix intervals.
----------------------------------------------------------------------------------------------------------------------------
I was skeptical about that itty bitty battery - (and antenna).  As the forum guys explained, it's all a matter of compromise.  iWatches are amazing at what they do. 

But certainly not good enough for zero to 34mph in a second or two.  No way Jose'.

PB's post piqued my interest, so I had a look around.

Location measurement via GPS without doppler is inherently inaccurate, both because of DOP issues (HDOP, PDOP, GDOP) and atmospheric issues. Fortunately, the feds turned off the time signal randomizers in 2000, so it's no longer wildly inaccurate, but it's still about +/- 5 meters. Increased sampling can help, but that doesn't change the simple fact that every measurement can be off by as much as +/-5 meters, and the distribution is somewhat random. You can be standing still and the GPS thinks you just jumped 10 meters in some fraction of a second. Raw GNSS data can be manipulated to give an accuracy of about +/- 1 meter. I don't know if GoPros, iWatches, Trace or any of the stuff you guys are using does that, but that's still a fairly shitty ruler.

Atmospheric conditions, line of sight, it's all in the Navigation engineering sites.  Consumer grade GPS for acceleration/speed/location is lame until you get into very specific, very expensive software.  That's how the DoD does it...  Big antennas, big power, double checking data.   
Support your local shaper

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #73 on: January 30, 2021, 03:06:42 PM »
Another example of a standard wave (2nd view).  The Blue Charts show the MAXIMUM and MINIMUM for every SECOND.  A Flatline would have dropped out of the the view. 
- Most watches on the market use 1-second GPS fix intervals

- The Apple Watch uses an interval that "varies." The definition of that is proprietary, but from what I've seen online (and anecdotally validated through using it) is that the the Apple Watch is taking a fix roughly every ten seconds then using an accelerometer to interpolate position. Apple has a high-res screen (big battery drain), so the reduced GPS fixes conserve battery life.

- The Apple Watch uses a generous GPS smoothing function. This makes really pretty lines on a map, but it isn't very responsive to pace changes, especially when combined with the increased GPS fix intervals.
----------------------------------------------------------------------------------------------------------------------------
I was skeptical about that itty bitty battery - (and antenna).  As the forum guys explained, it's all a matter of compromise.  iWatches are amazing at what they do. 

But certainly not good enough for zero to 34mph in a second or two.  No way Jose'.

On the one hand I should be happy someone even wants to pay attention to this.  On the other hand the responses are anecdotal.  The example given above, such as drop out, are for a single wave and virtually irrelevant to the issue of measuring speed.

The devices may not be perfect but are they within 1%, 2%, even 5%.  At 5% On a 20 mph wave that would amount to 1 mph difference.  1.5 mph on a 30 mph wave.  I'll take that.  That's just for speed.  Not to mention distance. 

DATA:
Here is a comparison between the TRACE which recorded the data at 5 intervals per second.  Not the GPS speed but the the records.  70,462 data points (extracted from Millions aka just the waves).  450 waves on a foil from 2017 to 2018.  aka Not just one wave.  See the MAX speeds.   

Note: The TRACE was likely more accurate than the iWatch.  However another way around any potential question on data is to overwhelm with data points.  With enough data the noise gives way to SIGNAL. 

As with artificial intelligence the algorithms haven't change what made it all possible was 1) Processor power and 2) more data. 

Did you bother to look up the Haversine formula.  When using Haversine all you need is GPS and Time.  The GPS accuracy is confirmed by tracking on land.   Millions of users giving feedback on Strava when the run isn't perfectly fitting on their path or sidewalk for example.  If they don't have it correct people scream.  The iWatch is on the right side of history.

2nd Example from the iWatch 14,700 data points.  Yes fewer but the data is FAR, FAR, FAR easier to access.  Similar Wave Count = 438.

See the comparison.  2 completely different technologies.  Max speeds, averages, distributions nearly identical.  NOT a single wave from almost 1,000 waves over 30 mph from either technology.  Why?  Because surf foils don't go that fast. 

White Screen - TRACE = Average Speed 12.9 mph

Black Screen - iWatch = Average Speed 13.2 mph
« Last Edit: January 30, 2021, 03:27:59 PM by Beasho »

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2843
    • View Profile
Re: Apple iWatch with Dawn Patrol SURF TRACKING
« Reply #74 on: January 30, 2021, 03:34:15 PM »
White Screen - TRACE = Average Speed 12.9 mph

Black Screen - iWatch = Average Speed 13.2 mph


This interesting because I have been using MORE modern wings lately.  Specifically the GoFoil NL Series vs KAI and IWA wings.

The other observation is the average distance went UP from 196 yards per wave to 241 yards per wave (from the samples).  MOST of the speed comes from the initial drop.  An average of 15 mph for the entire wave is VERY fast.  Meaning that the longer the wave the disproportionately greater time spent pumping on the slower tail end of the wave.

PS:  Attached are the 70,000 records from the TRACE in case anyone wants to dissect it further.  NO FLATLINES
« Last Edit: January 30, 2021, 03:37:16 PM by Beasho »

 


* Recent Topics

SimplePortal 2.3.7 © 2008-2021, SimplePortal