Contact   Imprint   Advertising   Guidelines

Trace errors compared to Woo

forum for kitesurfers


bmcfiv
Rare Poster
Posts: 34
Joined: Thu Aug 24, 2017 6:55 am
Kiting since: 2016
Local Beach: Sullivan's Island / Breach / IOP
Style: a messy freestyle
Gear: Eleveight FS09 / FS12 / RS 14 / CS bar
Zephyr Ozone 17
Xenon Infra 136
Ronix Boots
Mystic X
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby bmcfiv » Thu Jan 25, 2018 5:45 pm

great idea. we have a lighthouse in the water near by that i can use as a height reference. i think i can figure something out assuming i live after the jumps. hell yeah!

kiterocky
Frequent Poster
Posts: 405
Joined: Fri Apr 10, 2015 7:25 pm
Kiting since: 2000
Gear: Wipika
Brand Affiliation: Wipika

Re: Trace errors compared to Woo

Postby kiterocky » Thu Jan 25, 2018 9:58 pm

andylc wrote:
Thu Jan 25, 2018 5:35 pm
What might be more useful would be determining if ANY of them are accurate - have comparatory videos been done where height can be accurately measured and then compared to what the device thinks the height is?
Exactly....the producers have to do this...since than its just a toy...

andylc
Frequent Poster
Posts: 328
Joined: Fri Apr 07, 2017 3:26 pm
Kiting since: 2005
Local Beach: Exmouth
Favorite Beaches: Saunton Sands
Gear: Flysurfer Soul 8m, 12m
Lieuwe Shotgun 132
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby andylc » Thu Jan 25, 2018 10:45 pm

Given that they're based principally on accelerometers I doubt they will ever be particularly accurate. For instance how does an algorithm work out the difference between a C kite, which yanks you off the water but doesn't have long hang time, but if you get it right takes you pretty damn high, a bridled kite which doesn't lift as aggressively but floats you up potentially higher, a foil kite which seems to lift you quite gently off the water but then keeps going up and up and hangs you there for ages, what about those awesome jumps in sketchy conditions where you do a pretty good job but then get gusted way higher half way up...I have loved the fun of using a tracker device but it's really difficult to tell if the stats are at all accurate. I don't know enough about the algorithms they use to say whether different sorts of jumps can be interpreted differently of whether it's as simple as initial acceleration and direction. The Trace tracker tells you how long the jump was from takeoff to landing - again I'm not sure if this is based on position at takeoff and landing (ie GPS) or some other algorithm instead.

User avatar
deniska
Frequent Poster
Posts: 347
Joined: Fri Aug 30, 2013 3:40 pm
Kiting since: 2011
Local Beach: Plumb Beach
Favorite Beaches: Cabarete
Gear: Ozone, Core
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby deniska » Fri Jan 26, 2018 1:29 am

andylc wrote:
Thu Jan 25, 2018 10:45 pm
Given that they're based principally on accelerometers I doubt they will ever be particularly accurate.
check out ICBMs... guess how are they guided? ;-)
andylc wrote:
Thu Jan 25, 2018 10:45 pm
For instance how does an algorithm work out the difference between a C kite, which yanks you off the water but doesn't have long hang time, but if you get it right takes you pretty damn high, a bridled kite which doesn't lift as aggressively but floats you up potentially higher, a foil kite which seems to lift you quite gently off the water but then keeps going up and up and hangs you there for ages, what about those awesome jumps in sketchy conditions where you do a pretty good job but then get gusted way higher half way up...
get some physics and math classes. If you know acceleration curve, you can easily integrate the distance..
2 main sources of error:
1) consumer grade accelerometers on a chip are not super precise (unlike stuff used in rocket science)
2)it's tricky to determine precise take of time and vertical speed, especially when riding waves (it may not alway be 0 but the device may not "know" about it since there is too much "noise" when riding on water...

Well, the chips are getting better/cheaper (you can put 10 in one device and get the averages) and so are the algorithms..

fluidity
Medium Poster
Posts: 114
Joined: Sat Mar 19, 2016 11:20 pm
Kiting since: 2016
Local Beach: Ngati Toa, Plimmerton, Titahi Bay, Waikanae, Petone, Seatoun, Lyall Bay, Eastbourne, Lake Wairarapa
Style: Wave, jump
Gear: Old Flexfoil 9, Switch Element V4 11m, Ocean Rodeo Prodigy 7m 2014, Nitro14m v6, Nitro9m v6 12mSwitch Legacy2
DIY CAD ultra-concave wave twintips 1500 and 130mm with my own fin designs. Easy upwind, awesome carving.

Re: Trace errors compared to Woo

Postby fluidity » Fri Jan 26, 2018 10:15 am

andylc wrote:
Thu Jan 25, 2018 10:45 pm
Given that they're based principally on accelerometers I doubt they will ever be particularly accurate.
It's actually all quite technically feasible and practical these days.
Physics and maths, well established fields that have been solved since before slide rules were a thing.
Look at a chip like this: https://www.invensense.com/products/mot ... icm-20948/ and then imagine that chip's data being analysed hundreds or thousands of times per second, building graphs of 3 dimensional movement based on each direction of acceleration after correction for deviations due to rotation. Combine this with gaming industry software advances for the physics calculations for realtime 3D games.
I'm not a maths guru but I've dabbled. If you watch a delta printer at work it's a similiar sort of principle in reverse, the microcontroller is doing thousands of Pythagoras's theorem calculations per second to position the print head blazingly fast while 3 stepper motors are whizzing carriages up and down to keep the head at one level but moving in x and y axis directions. Most 3D printers use an Atmega128 chip which dates back to about 8 years ago but there are much faster chips around like the bluetooth variants of the ARM cortex M4 series (which is just one family of microcontrollers which would be ideal for something like a Woo).

bmcfiv
Rare Poster
Posts: 34
Joined: Thu Aug 24, 2017 6:55 am
Kiting since: 2016
Local Beach: Sullivan's Island / Breach / IOP
Style: a messy freestyle
Gear: Eleveight FS09 / FS12 / RS 14 / CS bar
Zephyr Ozone 17
Xenon Infra 136
Ronix Boots
Mystic X
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby bmcfiv » Fri Jan 26, 2018 10:23 am

fluidity wrote:
Fri Jan 26, 2018 10:15 am
andylc wrote:
Thu Jan 25, 2018 10:45 pm
Given that they're based principally on accelerometers I doubt they will ever be particularly accurate.
It's actually all quite technically feasible and practical these days.
Physics and maths, well established fields that have been solved since before slide rules were a thing.
Look at a chip like this: https://www.invensense.com/products/mot ... icm-20948/ and then imagine that chip's data being analysed hundreds or thousands of times per second, building graphs of 3 dimensional movement based on each direction of acceleration after correction for deviations due to rotation. Combine this with gaming industry software advances for the physics calculations for realtime 3D games.
I'm not a maths guru but I've dabbled. If you watch a delta printer at work it's a similiar sort of principle in reverse, the microcontroller is doing thousands of Pythagoras's theorem calculations per second to position the print head blazingly fast while 3 stepper motors are whizzing carriages up and down to keep the head at one level but moving in x and y axis directions. Most 3D printers use an Atmega128 chip which dates back to about 8 years ago but there are much faster chips around like the bluetooth variants of the ARM cortex M4 series (which is just one family of microcontrollers which would be ideal for something like a Woo).
Incredible... I just checked out the price, $5!!!! Separately, any chance you could build me a laser bazooka for the front of my board?

andylc
Frequent Poster
Posts: 328
Joined: Fri Apr 07, 2017 3:26 pm
Kiting since: 2005
Local Beach: Exmouth
Favorite Beaches: Saunton Sands
Gear: Flysurfer Soul 8m, 12m
Lieuwe Shotgun 132
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby andylc » Fri Jan 26, 2018 12:46 pm

I suppose I was wondering if the accelerometer is just measuring the takeoff or continuing to analyse right up until the top of the jump, in which case you’d think it would be fairly accurate. I can’t imagine any of the consumer devices have particularly high grade electronics in them.
No thanks to the physics and maths classes though - I’m not that bothered!

kiterocky
Frequent Poster
Posts: 405
Joined: Fri Apr 10, 2015 7:25 pm
Kiting since: 2000
Gear: Wipika
Brand Affiliation: Wipika

Re: Trace errors compared to Woo

Postby kiterocky » Fri Jan 26, 2018 1:39 pm

Even the plane accellerometer are constantly correct by gps and computer....and they use the expensive one...😁😁 becouse even the most expensive can have 100% accuracy...imagine your china 5$ accellerometer 😁😁😁

User avatar
deniska
Frequent Poster
Posts: 347
Joined: Fri Aug 30, 2013 3:40 pm
Kiting since: 2011
Local Beach: Plumb Beach
Favorite Beaches: Cabarete
Gear: Ozone, Core
Brand Affiliation: None

Re: Trace errors compared to Woo

Postby deniska » Fri Jan 26, 2018 3:43 pm

kiterocky wrote:
Fri Jan 26, 2018 1:39 pm
Even the plane accellerometer are constantly correct by gps and computer....and they use the expensive one...😁😁 becouse even the most expensive can have 100% accuracy...imagine your china 5$ accellerometer 😁😁😁
All inertial navigation systems suffer from "integration drift" (i.e. small measurement errors build up over a long period of time)
So it does not work great on planes..
powered phase in ICBM only lasts a few minutes so those things get incredible accuracy for the range with inertial guidance.. figure 500m for 10,000km.
In kitesurfing, you would only need to measure 3-4 seconds each way, so with good accelerometer you can probably get it with cm precision...

kiterocky
Frequent Poster
Posts: 405
Joined: Fri Apr 10, 2015 7:25 pm
Kiting since: 2000
Gear: Wipika
Brand Affiliation: Wipika

Re: Trace errors compared to Woo

Postby kiterocky » Fri Jan 26, 2018 7:17 pm

Good accellerator 5$....😁😁😁😁


Return to “Kitesurfing”

Who is online

Users browsing this forum: Bing [Bot], decay, Google [Bot], Jwoo and 23 guests