Measuring Turning Force other data...

[st_lupo]

Hi all,

 

One of the things that blew me away when doing my reading on BOL and the Carver's Almanac was the point that really good riders on GS gear are pulling between 2 and 3 g's in some turns.  This is my first season so I know I'm nowhere near those extremes but I thought it could be interesting to measure what I'm actually getting out of the board and maybe track my turning force over the course of a season and see if it can be used as an indicator of my progress.

 

I've got lots of multicopter kit laying around the house and was thinking I could reprogram one of the smaller boards to log the accelerometer/gyro data to an SD card and then mount the whole thing on my board, in between my boots.  I guess there is a bunch of other information in there that could be interesting: roll rates during transitions, may even be possible to extract turn radius and skidding.  And on the cheap.

 

The question is has anybody already done something similar and maybe has some insight into what the big challenges are going to be related to application on a snowboard?  Programming and soldering iron is no problem, I'm thinking mostly along the lines of longevity of the sensors when exposed to snowboard vibrations, feasibility of integrating the recorded (filtered) data to generate a navigation history based on inertial measurements, etc.?

 

[Keenan]

Yes, they sell them on this very site:

 

http://bomberonline.3dcartstores.com/G-Lab-Meter_p_149.html

[jng]

Has anyone tried an iPhone app for charting acceleration?  

[Calle]

I tried logging it with my phone. With the phone in the pocket of the jacket theres probably many soruces of error. But anyway...

Carving a 19 m radius board at 50 km/h showed 2 G on my phone. Could not find any good app that logged the data and showed nice graphs for my windows phone.

[st_lupo]

One of the reasons I was thinking of working on a solution that moves the sensors to the board is to decouple the rider's movements from the measured values.  Using an iphone app I can stand in place and hp and log 4g's.  Additionally the rider's body movements/rotations will influence the accelerations that the sensors see, and contributes to error in measured value (imagine holding the phone in your hand and waving your hand around, these measured accelerations are not necessarily applied to the board). 

 

The weakness with the Glab device is that it reports a single max g-acceleration.  Without better insight into the filtering that is applied to the data to remove shocks, this acceleration could maybe be the decelleration during a hockey stop instead of your best turn.  I'm thinking of trying to making an easily built system that tries to extract a more comprehensive set of telemetry over the span of a day and is useful/interesting for race training or obsessive engineers (but mainly because I think building these kinds of things is fun).

 

Centralizing the sensors on the board should allow a reasonable estimate of instantaneus turn radii, amount of skid, etc.  Being able to plot the various data elements in a strip-chart format next to each other (and maybe a visualiation of the board movements) should allow evaluation of each turn during a run or a whole day.  Hopefully this could assist objective evaluation of performance of both the rider and the board.  For example you might evaluate your ability to maintain a constant turn radius through complete turns, and see if you need to focus on the initiation, middle or termination of the turn.   Or quantitatively compare the performance of two different boards.  To get a reading of how much force is being transmitted from the board to the snow it might be possible to mount transducers under the bindings that measure the force exerted by the rider to the board. 

 

I think I've got one last day of riding this year so I'll try to wire something up and record the sensor measurements, then spend a few hours this summer to figure out how to post-process it. 

[bobble]

i had thought about doing such a project.

 

i had bought an accelerometer board from sparkfun that measured x,y,z axis. i went driving around the neighborhood to take some measurements. when i got back i looked at the data and it was extremely noisy. i was surprised. very choppy. i used a kalman filter  to smooth out the data using MS Excel. i had a hard time interpreting what i saw as the data was still somewhat noisy. kind of strange since i wasn't driving aggressively. just a few quick starts and brakes. maybe the board was bad. i dunno. maths are my strong suit. just like to build shit.

 

thought about how to mount the device. some place between the bindings but kept thinking that there would be more noise from riding on a snowboard than a car.

 

there are other issues to contend with such as zero-ing out the sensors before the start of a run. the sensors have drift so they require compensation. i think the ambient temperature affected the sensors as well.  thought about using an Arduino but then i just lost interest. i think i started snowboarding or something.

Edited April 7, 2015 by bobble

[Corey]

There was a big discussion about this a few years ago, but I can't find the thread.

I have an earlier version of this data logger for car racing, but I'm concerned that a little chatter would destroy it. They void the warranty on those units if used on race karts, and the shock loads in a snowboard are MUCH higher as there are no air-filled tires soaking up ground vibrations.

My desire for data is outweighed by the cost. Plus, like Bobble, I quickly get to the point where I'd rather just ride.

[Neil Gendzwill]

If you can hold a 10 m pure carve at 30 km/hour, that's around .7 G.  You'd have to be hanging on to a 10 m turn at over 60 km/hour for a 3G sustained turn, which I don't think I've seen anyone do.  You might measure instantaneous forces of 3 G but I highly doubt people are generating sustained lateral acceleration that high.

Edited April 7, 2015 by Neil Gendzwill

[jng]

 Hopefully this could assist objective evaluation of performance of both the rider and the board.  For example you might evaluate your ability to maintain a constant turn radius through complete turns, and see if you need to focus on the initiation, middle or termination of the turn.   Or quantitatively compare the performance of two different boards.  To get a reading of how much force is being transmitted from the board to the snow it might be possible to mount transducers under the bindings that measure the force exerted by the rider to the board. 

 

 

I think you would find yourself disappointed that you will not be able to keep all other variables constant, e.g. snow conditions, turn shape, slope, etc.  Better to focus on your body inputs via video analysis than the board outputs.

 

I would be satisfied with an app since all I plan to do is pull the x/y/z components and calculate scalar values strictly for amusement purposes.  There will be noise from body movement during edge transitions, but max G's are between transitions.

[bobble]

i'll just leave this here...

http://www.vernier.com/innovate/accelerations-in-snowboarding/

[Corey]

Bobble - that's cool! It's an ad for that equipment, but still interesting. Was that a guy on softboots pulling 2.5 Gs? Impressive!

st_lupo - can you get pictures or videos of yourself for your own self-critique? Lean angle is a direct indicator of lateral G's, at least until body parts start touching the snow*. The hard part is figuring out where your center of mass is, but you're probably just looking at rough estimates anyway.

* You can see my avatar where my hand is on the snow. It's not doing much, probably bearing 2-5 lbs of weight. (For the record; this is a risky behavior. I hurt my shoulder on the first day of SES because of this.) If you do the math, it's surprising how little force is needed to allow considerably more lean angle! See the truly-gifted EC riders that make it look effortless. As they get closer to truly laying on the snow, the force their board supports approaches infinity. But they support some weight with their bodies, as evidenced by the required clothing reinforcements, which allows them to get impossibly low.

[Beckmann AG]

The question is has anybody already done something similar...

 

Yes, but I'm not at liberty to discuss it. :(

 

 

...and maybe has some insight into what the big challenges are going to be related to application on a snowboard?

 

 

The issue isn't so much the hardware, it's what you plan to do with the data.  

G readings are good for barstool banter, but that won't do much to guide your progress as a rider.  Similarly, the data you collect should tell you something about 'what is'; but then how do you plan to get to 'what should be'?

 

Given that your board responds primarily to tilt and variable loading along it's length (both of which can be resolved in terms of 'pressure'), you might consider sampling load change under each foot, and also at the toe and heel of each foot.

 So at least six collection points.

 

A number of years ago, someone had available a pressure sensitive insole, presumably to help in some variant of the 'balancing and alignment' scheme.  From the description at the time, it didn't amount to much more than smoke and mirrors, but in the right hands a tool of that type would be useful.

 

If you want to make a marketable product, come up with something that quantifies hand location and proximity of the CM to the top-sheet and/or the snow. ;)

[Corey]

A number of years ago, someone had available a pressure sensitive insole, presumably to help in some variant of the 'balancing and alignment' scheme.  From the description at the time, it didn't amount to much more than smoke and mirrors, but in the right hands a tool of that type would be useful.

I worked on exactly what you describe 10-15 years ago at Vista Medical. http://www.pressuremapping.com/ It was part of a very neat system, designed as a tool for multiple theraputic uses. We had only developed one size of insole when I left, but it indeed measured foot forces to a surprising level of accuracy and repeatability even at low values. The same technology could measure fingertip pressures on a mattress, for reference. The biggest concern was the thickness, but they've probably made large strides (ha!) since then.

Combine that with the strips around your calf that we'd put in a leg prosthetic socket and you'd have some neat data. Again though, what do you do with that data?

These guys do similar: https://www.tekscan.com/

Two custom sensors under the E-ring of a TD3 would paint a VERY interesting picture. They would resolve all the normal forces that the rider puts into the board.

[bobble]

Two custom sensors under the E-ring of a TD3 would paint a VERY interesting picture. They would resolve all the normal forces that the rider puts into the board.

Thought about that too. Saw this awhile ago.

http://www.researchgate.net/profile/Florian_Michahelles2/publication/3437104_Sensing_and_monitoring_professional_skiers/links/0046351ff83c1c99e4000000.pdf

[Beckmann AG]

^Seen anything recent relating to that study?

[st_lupo]

That "Sensing and Monitoring..." paper is really interesting.  Some points are essentially what I've been envisioning, but maybe on a smaller/cheaper scale. I like the idea of the IR inclination sensor and might just have to try that out.

The plan so far is:

--This/next week

*Build/program/calibrate a prototype system out of common household items that only logs 6-axis IMU data. 

*Collect some data next weekend.  This time it is definitely positively absolutely my last ride of the season.

--Over summer

*Spend some time implementing filtering and navigation algorithms to estimate velocity and position based on logged data

*See what other interesting data can be extracted.

*Graphical presentation of the data

*Do some dry-land tests to quantify the errors that are introduced due to sensor drift and filtering.  (Since my home mountain is relatively small each run is around 3 minutes and hopefully sensor drift won't impact too heavily)

--Future

*Telemetry of force from bindings.  Isolate each binding to three contact points and measure forces.  Decompose to 1 normal force and 2 orthoganal moments (coincident to the contact plane).  This will give us fwd/aft weighting and twist/bend that the rider is applying to the board.

*Better IMU, might be able to borrow one from work to compare with the initial low cost system

*GPS, kalman filtering?

 

I did some 2D partical tracking analysis using video for some friends a while back.  Could look into lowcost 3D that would fit on a board :)

 

I guess if I run into too many difficulties I'll pack up everything and just have a bike and a beer instead.  Either way it's kind of a fun alternative to sleeping and I learn a little more about inertial navigation which kinda helps with my day job...

Edited April 9, 2015 by st_lupo

[philw]

The Vernier thing looks handy, but as he says in the text it's quite noisy data generally, so you need some interpretation.

 

Depending on what you're trying to achieve, you may find it more useful to look at it the other way around - look at your track, and work out what the force required for that would be.

 

... Was that a guy on softboots pulling 2.5 Gs? Impressive!

 

I'm sure it's not what you mean, but if you're going to hold a curve then the force certainly doesn't know what you're wrapping your feet in.

[bobble]

@Beckman... No I have not seen anything related to the study. I liked the concept of using FSRs on the boots as it seemed fairly straight forward. Instead of boots I thought mounting the FSRs under the bindings. I had envisioned using an Arduino as the data logger but after doing research the sample-and-hold circuit is multiplexed -- the force measurement would not be entirely accurate from one input to the next at the speeds we travel. Also the A/D conversion is not fast when you scan all analog inputs. The packaging and cabling of the device discouraged me from pursuing the project further. 

 

@st_lupo... what are you using for your data logger?

[st_lupo]

Hi Bobble,

I'm using an arduino 328p for the inertial-only prototype.  The biggest reason that I'm using this is that I had a spare one in my stack of multicopter flight controllers and a MEMS IMU is already integrated via I2C on board.  I wired up a micro SD card port and I think (hope) I should be logging data at around 200 Hz.  In a previous job the navigation, guidance and control routines ran fine at 256 Hz and that was a bit more of an energetic application than this.  

 

Tonight I quickly calibrated the gyros and ran a quick test; the total RSS error in the orientation quaternion after 3 minutes is around .3 degrees when sitting stationary, so it doesn't look too bad.  If/when adding additional sensors I would probably look at one of the newer/faster 32bit arduinos or one of the new raspberry pi 2s.  The nicest thing so far is that the total hardware expenditure for this first prototype, excluding battery and sd card is around $25. 

[RCrobar]

This Is What Happens When Scientists Go Surfing

Hello

I have zero background in this area but did stumble across this surf video that seems directly related to the discussion, thought you guys might find it interesting.

Cheers
Rob

[breeseomatic]

TLDR;

 

This is what we have in our track car:

 

iPhone 5s for logging & basic accelerometer (free)

and running Harry's Lap Timer app ($25)

SkyPro XGPS160 GPS Receiver. ($150)

 

Run any combination of free apps and you can probably get some decent data.

 

Securely mounting to a snowboard directly could be a problem both in accurate data aquisition and electronics retention and reliability, I recommend using an isolation plate.

 

Corey, you may like this for your autocross sessions.

Edited April 10, 2015 by breeseomatic

[Corey]

Corey, you may like this for your autocross sessions.

A friend uses Harry's on his iPhone. It's pretty decent. I've tasted the good stuff, Race Tech DL1 and also SoloStorm on an Android tablet. They're dramatically better for autocross! But they're not $25... I think SoloStorm might be the most expensive app ever at $200, but it's still the one I've received the most value out of.