Pylon Tracking

[tap]

It's been about six months since I've made any progress with pylon tracking data, so it seemed like a good time to make my ski partners suffer through another day on the lake with computers, wires, and sensors (they are very kind! @Gloersen ). The approach I took before was to put strain gauges directly on the pylon and then extract load magnitude and load direction from pylon deflection. It worked pretty well, except when the load was too low the signal to noise ratio got pretty rough. So, I modified the setup and am now taking load data directly in-line with the rope and rope angle directly as well with a little swivel tracker. I'm pretty happy with the results. The swivel tracker had a few degrees of mechanical play in it, so I need to snug that up, but the data seems good enough to share.

 

Here's some data graphed out (slightly low pass filtered).

 

 

 

There is a lot you can do with this data; I find the hardest thing is to figure out how best to present it. It seems like a good place to start is simply to plot out the physical handle path, so here you go...

 

 

Keep in mind the few degrees of slop in the swivel hardware, which explains why it seems to come up slightly short at the turn balls. I tried to scale the graph to the correct aspect ratio to make it visually correct. It's pretty much a saw tooth pattern. It will be interesting to overlay different line lengths and speeds.

 

I had a few requests the last time around for a copy of the raw data, which I pretty much ignored (sorry about that). The data was just not clean enough at the time to really let it out. It's still not where I want it, however it's probably good enough that if anyone wants a copy just send me a message. I'd be happy to contribute to the wasting of free time.

[PurdueSkier]

Cool stuff @tap. Nice work putting that Purdue degree to good use.

[Alberto Soares]

@tap - Very nice, keep it up. Is the skier LFF?

[tap]

The skier happens to be me this time. Right foot forward.

[tap]

Another night, another graph. This time I've added Handle/Skier Speed in mph. The speed is absolute speed relative to the surface of the earth. Speed is the dotted red line overlaid on top of the handle path through the course. For this pass I seem to peak right about 45 mph somewhere around the whitewash on the way out to buoy line.

 

Personally I don't think absolute speed tells you much of anything useful. It's simply too dominated by the transnational speed of moving down course. But... I hear people ask the question quite a bit, so here it is.

 

 

I get much more use out of angular speed relative to the boat. And no, they do not peak in the same place. Angular speed peaks much sooner, well before center line if things are going well.

 

[tap]

Another night, another graph. This time I've added Skier/Handle Speed overlaid on top of handle path through the course. Speed is absolute speed relative to the surface of the earth in miles per hour. Personally I like thinking in terms of angular speed about the pylon much better, but I hear this question asked quite a bit... so here it is. Speed is the red dotted line and uses the scale on the right hand side of the graph.

 

 

I seem to peak right about 45 mph somewhere around the whitewash on the way out to buoy line. A point of interest, your angular speed peaks much sooner, quite a bit before center line if things are going well.

 

[Gloersen]

@tap is going to gather some very cool data! Once he gathers enough data from expert short-line skiers, then to overlay with skier data struggling to attain short line passes consistently; the aspiring skier will have a better understanding of implementing technique with positive results.

 

In other words, when a skier is working on improved technique, this type of data can indicate if results are in a positive direction. Some might argue that running more buoys is indicative of improving technique, but for the many overly analytical skiers and those that have hit a “wall”, this data will be awesome. It’s great to have talented and resourceful skiers like @tap with a passion for this sport.

 

[DefectiveDave]

@tap,

 

Very cool and thanks for all the hard work! Are you accepting donations to continue this work? I would gladly donate a little bit if there was some mechanism to do so. Probably won't make or break the project, but it could be a good way of showing appreciation.

 

Also, just out of curiosity given the current data, are you doing a 2 handed gate with a glide? What line lengths do you normally ski?

 

I'm already trying to learn something from the data, haha.

[Wish]

@tap @Gloersen Thanks again for allowing me some sets with you all. I regret not having you gauge my skiing and mapping it out. Would be cool to see it now that I kinda understand it a bit more. I think much if not more thanks goes to @tap wife for putting up with the extra time to do such a ambitious project. Very cool to see in person.

[tap]

@DefectiveDave No means for donations, but thanks for the offer. Correct, 2 handed gate with a glide. As far as my ability, I'm just an average skier but on a good day I can scrap through 32 off at 36 mph.

 

I fully intended to video the passes that I measured, but somehow completely overlooked the obvious and didn't get the camera out. So this is not the measured pass, but from a set the week before, an opening pass at 28off/34mph. My bad habits are ever present, so it's probably a reasonable representation of what was measured.

 

 

[tap]

So this is kind of interesting... I measured a 28 off pass at 34 mph and 36 mph in back to back passes. I overlaid the handle path and the speed of both. The handle path is nearly identical, at least without zooming way in. I had to double check to make sure I was pulling the right set of data. So at least for me, I run a pretty similar path just at higher speeds. Peak speed at -28/36mph was 48 miles per hour.

 

 

[Deke]

@tap how did you get the speed data? Is that calculated from the angular velocity? Just curious, thanks.

[tap]

@Deke exactly... well mostly anyhow. I actually did it two different ways to see which was cleaner and as a means to double check my own work. The first was the angular velocity approach... convert angular velocity at each time step to a Cartesian velocity vector (i.e. centripetal velocity), then add the velocity vector of the boat to change the reference frame from the boat to the lake. The alternative approach was to map out the Cartesian coordinates of the skier (handle) at each time step with reference to the lake, then take the discrete derivative per each time step to get the magnitude of the velocity vector. The results from the two methods overlaid pretty well, but the low pass filter seemed to do a better job with the first approach. The filter just takes out some of the noise in the data. The trick is knowing how much and when to apply. The reality is it smooths out some of the peaks, so when the charts shows 48 mph as a peak velocity, the unfiltered data actually showed 51 mph. However, that 51 was such a finite peak that it was most likely noise. And if it happened to be real it certainly wasn't sustained for any duration worth mentioning.

 

I think I just made that sound much more complex than it really is. All you really need is one physics equation and some vector math.

[disland]

I want to see Nate Smith on that thing

[disland]

I find it very interesting that you are slowing down so much before turning in for the gates. It would be interesting to see on video if your falling back relative to the boat. It would also be interesting to see if the same thing happens at 32 and 35 off

[DefectiveDave]

@tap,

 

How much custom fabrication is required to make your setup? Like, could I just buy all the hardware, flash your software, and strap it to my pylon?

 

Also, how did you attach your load cell to the rope? I already have one of these, but I was a bit hesitate to attach it without a large radius of curvature piece and that proved challenging at the time.

[Triplett]

Hmm.. What is interesting is the difference in handle path on either side. Keep it up, this is some really cool data you have here.

[ScottScott]

I noticed what @Triplett said also. Would be interesting to flip those over (and shift down a ball) and overlay the 2 sides on top of each other. To me it looks like there's a much nicer arc on your offside than the onside. I think this is something that happens often with onside. We feel more comfortable on our onside but don't realize that its so easy to open up toward the boat and loose some angle (getting pulled downcourse,) taking a straighter line into the ball into your offside turn. This is also supported with the speed graph showing that you slow down into the ball better on the offside. I think we often have difficulty on our offside turns becouse of our onside cross. I was surprised when I started getting some coaching with Jodi Fisher and his focus has been my alignment on my onside turn and cross.

 

Bottom line....this is a great training tool, with lots of information to be gathered from these graphs. This should be produced into a tool that can be easily attached and uploaded into a tablet. Maybe just for ski school application, and the occasional personal use. I need one.

[tap]

@Mark_Matis precisely.

 

@disland I definitely tend to drift in as I approach the gates (despite my best efforts), so I'm not too surprised that I'm slowing down quite a bit.

 

@Triplett @ScottScott yeah I agree, there's definitely some imbalance there. Here's a graph of the -28/36mph pass with the handle path inverted and shifted one ball to the right. You might have to zoom in a bit, but there's definitely a deference from one side to the other.

 

 

Here's the same handle path (-28/36mph) but with the rope load plotted along with it. I think it's really interesting to see the difference in how I'm working the boat in my heel side vs. toe side pull. It almost looks like I double pull at the wakes from 1-to-2, 3-to-4, and 5-to-6. It's crazy how consistent the imbalance is. It's almost like a flashing neon light identifying something that needs to be fixed. I just need to figure out how to fix it.

 

 

[tap]

@DefectiveDave my system is not exactly elegant, but it is quite functional. The setup is reasonably straight forward. A little bit of soldering, a little bit of fabrication, and a lot of time playing with Excel (our your number crunching software of choice). I'm not a software guy, so I generally stick to what I know. The whole system really hinges on the use of a data acquisition system (DAQ). There's several companies that make these that provide commercially off-the-shelf introduction models for the home user. There's nothing special about the collection side of the software, whatever comes with the DAQ is probably sufficient. Your restrictions with the DAQ are sampling rate and finding one with a built-in instrumentation amp. Load cells function using a wheatstone bridge, hence the need for the amp. If this is all making sense then you're probably good to go, if not then I'd suggest doing a little more homework before throwing any money at it. If you're really up for giving it a go send me a message and I'll fill in some of the details. Expect to throw away a few hundred bucks, several nights of computer time, a couple nights of garage fabrication, and lots of strange looks as you show up to the dock with a laptop and a mess of wires.

 

As far as the load cell goes, make sure you're using one with sufficient load rating including a decent safety factor. I've measured skiers near 700 lbf of load (although he's a bit of a freak). I'm using a simple S-type load cell with some beefy load rated eye bolts at each end. I've made it the same length as the 41' off loop, so I just take that loop out of my rope. I'm not concerned about the tight radius of the rope at the connection since this is typical of how ropes are connected from one loop to the next. Load tested everything, did great.