chill guys I don't doubt these guys can jump high respect for just launching an 8m in 48knots (what he says it was blowing on his FB page)
All I'm saying from my experience using woo and riding with friends that have it woo is just a nice reference to compare height with friends just not even close to reality.
If you want to prove a 25M jump put a kite on 25m lines at 12 and jump over it and touch the canopy post the video then we believe
Woo is IMU (inertia motion unit) which uses electric sensors to measure jump. 3 axis Gyroscope to detect rotation and 3 axis accelerometer to detect acceleration on different axis. Also clock is used (hangtime clock).
First unit rotates coordinate axis to earth and calculates accelerations by earth coordinates. Sensor can be oriented any direction like board inverted. It still calculates direction to center of earth. Then it evaluates on that axis forces up and down to get acceleration component for height. (Unit also does other signal processing like filtering sensor inputs etc)
Unit integrates over time the acceleration to get speed. Speed is integrated again over time to get position. Position is the woo-meter height.
Like basic physics formula
y = y0 + vy*t + 0.5*ay*t*t
but here the sensor signal is constantly changing every 0.001s (if 1000hz sampling). But actual formula is double integrated over time to get area of acceleration to get the speed and then area of speed to get distance (height).
This is why the hangtime must be correct because it is evaluated on speed and then on position calculation. And checking the hangtime from video against leaderboard is the minimum requirement.
Using accelerometers to calculate position (height in the case of what woo tries to calculate) is loaded with large errors. Accelerometers experience random noise and the readings drift with time. When you double integrate over acceleration to find the position the drift error grows quadratically. Great short video showing why tracking position based on accelerometer data doesn't work:
Swimmer, I wonder why your linked video does not use Madgwick AHRS filter or does it? This video shows better results in terms of drift compared to your linked video?
tracking position based on accelerometer data doesn't work
Except for all those submarines out there that have been using inertial navigation for about 70 years...
Kiteboards and submarines aren't the same, but inertial navigation is one of the core backbones of tactical military navigation systems that buttress GPS during periods of interrupted service.
Swimmer, I wonder why your linked video does not use Madgwick AHRS filter or does it? This video shows better results in terms of drift compared to your linked video?
Yes you can absolutely do some sort of filtering (Kalman filtering) to reduce drift but you have to know the approximate profile of the motion or use another sensor (sensor fusion) to essentially do a weighted average for a better estimate.
3D foot tracking may work a little better as each step is under 1sec in most cases, so the amount of drift that can happen with each step small but over many steps you will still accumulate some error. Also the step motion has a somewhat predetermined profile so you can use filtering to get a decent estimate.
Now estimating the profile of a kiteboaring jump that can last greater than 2sec is much more complex!
There is a reason why Woo doesn't state the accuracy of it's devices
I don't doubt that people have jumped 25m or more especially in capetown conditions it's just Woo numbers are a bit meaningless.
Except for all those submarines out there that have been using inertial navigation for about 70 years...
Kiteboards and submarines aren't the same, but inertial navigation is one of the core backbones of tactical military navigation systems that buttress GPS during periods of interrupted service.
Except subs don't just use accelerometer data and double integrate it to find position
In conjuncture they get information from gyrocompass, measured speed, speed of local ocean currents, bottom contour navigation, etc
Well, I would believe that Olly Bridge did a 30m jump at Kitebeach on Bloubergstrand in Cape Town today. He was on a 8m Flysurfer Soul and ripped a jump that for all appearances was way above a typical kite at apex. Hangtime was ridiculous, his helicopter loop at the end took him back up again for a bit, and as he landed he hit another wave and did another jump. Pretty impressive stuff. Lots of hotshots on 9m kites ripping big jumps and megaloops, like local guys Josh Emmanuel on Core and Stuart Downey on Naish. Also one older guy I don't know on a 9m Airrush Vantage 3, which is an unusually high aspect kite for a LEI, sort of inbetween typical LEI and foil. The recorded wind was 25-35 knots, see below, but I could barely hold on to my 7m Pivot. It was what the locals call a "black SEaster" with dark grey clouds coming through from Muizenberg and Table Mountain covered with a blanket of cloud rather than just a table cloth.
Attachments
Tuesday January 8 2019.jpeg (88.55 KiB) Viewed 1170 times
Yes you can absolutely do some sort of filtering (Kalman filtering) to reduce drift but you have to know the approximate profile of the motion or use another sensor (sensor fusion) to essentially do a weighted average for a better estimate.
3D foot tracking may work a little better as each step is under 1sec in most cases, so the amount of drift that can happen with each step small but over many steps you will still accumulate some error. Also the step motion has a somewhat predetermined profile so you can use filtering to get a decent estimate.
Now estimating the profile of a kiteboaring jump that can last greater than 2sec is much more complex!
There is a reason why Woo doesn't state the accuracy of it's devices
I don't doubt that people have jumped 25m or more especially in capetown conditions it's just Woo numbers are a bit meaningless.
it's not all doom and gloom as you presented:
- you don't care about 3D position, only vertical component... so less error from the drift
- average jump is only about 3 seconds in each direction (you reset once your Z vector changes direction, measure height one more time on the way down and average).. so less error.
- you can put more accelerometer chips for better filtering.. they are small and cheap (only a few bucks a piece).