Trace / The Action Sports Tracker

[Horton]

It is early days using Trace for water skiing but after spending a few hours with David Lokshin yesterday I am pretty fired up.

 

Some of the speeds in the below videos looks wrong. There is a .3 second delay between some the video and the speed data. Again this is early days but how cool is this?

 

There is not much delay on the first video. The second one down has some issues. I will say it again => this is early days

 

http://youtu.be/GAnXL5pfrT8

http://youtu.be/kCCoFN5fqMo

http://youtu.be/nvmP5qFZ7k0

[bbirlew]

Very cool. Was the data split up into passes automatically or manually?

[Mateo_Vargas]

Interesting that your highest speeds are on your gate.

[Horton]

@bbirlew haaa I don't know. I am still scratching the surface on this stuff. Lots of questions but I am super fired up.

[Zman]

Are vids 1 and 3 tracking speed in direction of travel; vid 2 speed in direction of boat?

[mwetskier]

assuming the helmet is designed to keep the alien mother ship from reading your thoughts do you have to provide the tin foil liner your self or does it come with that?

[Horton]

@zman no. I think in the second video the sync between speed data and video are off. All of the speed data is from the top of my head - literally

[Steven_Haines]

Cool stuff!

[ktm300]

I love this. Data rather than hypothesizing. The first time I rode in the boat with a strain gauge connected to a lap top tracking the load points, I was really surprised. With one of the smoothest 34 skiers on earth behind the boat, peak load was 650ish (lbs) within a ski length inside the buoy line. Dropped to around 400-450 by the centerline. No matter what we think, the highest load is out at the buoy line. All the talk of speed through the turn will get tested with this device. In video one, your speed at/near the apex is about 10mph higher offside than onside. As a lifelong digger of holes on my onside, that explains it right there. I always thought I was turning too hard, wrong, etc ... I bet I am just slow and literally skiing out of a hole.

[ktm300]

Oh yeah, turn in point too. Slower. I am not criticizing anything... just recognizing my own shortcomings with regard to my onside turn.

[MISkier]

Your peak speed during the pull from your onside turn to your offside looked like it was 4-5 mph faster than the pull from offside to onside - at least in the 3rd video.

 

It looks like it will be an interesting analysis tool.

[Deanoski]

John what # are you running on the Vapor fin and boot? is it a 67?

[dchristman]

We're going to have to get two Tracer's - one for the handle and one for the ski. Maybe they can come up with a mount that fits the Y of the handle. It would be interesting to see the two simultaneous tracks plotted on a course map with color coded speed.

[dchristman]

@horton can you post the data files for these?

[dchristman]

I have been able to determine from the data @horton previously provided that he skis very erratically and lives in paradise! That's a dozen freakin' lakes :D This must be the promised land!

 

 

[Horton]

@dchristman‌ the above data trace is funny but is from my first try and was bad.

[tap]

@Horton‌ , very cool. Any idea what the sampling rate is?

[Horton]

@tap not sure. It is not only gps. It is also accelerometer data (I think). So that is me saying I do not know

 

@rico‌ somehow it is automatic. Cool but explains the inconsistency. I have asked for more information

 

 

[Alberto Soares]

I want one!!!

[skiep]

Is there one in the boat too?? It's hard for me to belive the boat is moving that much

[skiep]

Ok - My bad I guess that is you're path upper left

[Horton]

Here is a good path

http://surf.traceup.com/stats/u?uId=380303&vId=404

[MattP]

 

[MattP]

 

[disland]

Would be interesting to take a run strait down the lake and see how close the speed is to zero off

[Horton]

@disland‌ I'll do that this weekend

[dlokshin]

Hi All,

 

I'm David, one of the co-founders of Trace. We never really thought about using Trace for water skiing, but John's initial issues with accuracy was a technical riddle we couldn't figure out (now solved) and couldn't pass up (I love riddles!). With that settled his enthusiasm is a little contagious and my curiosity is piqued!

 

Great to see the feedback on this forum. I just wanted to stop by and offer to answer some questions if you have any. I can't promise I'll be a regular here, but I'll do my best to check back. Not quite sure how these forums work, but if you @dlokshin me when you have a specific question, and I get an email notification, that would be super.

 

Thanks!

[brooks]

This product is super rad! It is something I had hoped to see for a long time and to see the possibility of it being very feasible is really cool. To be able to track path and speed adds another element that for forever has only been dreamt about. I am pumped by your enthusiasm @dlokshin and hope that this product can be readily available to skiers as it would be a huge tool for the ski world! It is definitely something I know K.C. and I would use often. Excited to see what comes of this.

[AdamCord]

I MUST have this! Super Cool.

 

@dlokshin here are my questions:

How is the video synced up with the data like that? Does your software do that if I have a video file of the pass?

What is the sample rate of the data? We accelerate/decelerate quickly so wondering what the lag is and time between data points.

Does this unit incorporate 3 axis sensors and accelerometers? It would be even more useful to have acceleration data as well as pitch/roll/yaw.

Is there any way to make it work mounted to the ski? I know John initially had problems with losing signal when it went under water. If it's on the tip of the ski it could feasibly be kept out of the water at all times.

 

Thanks in advance, I hope you guys sell a few million of these things.

[disland]

@dlokshin Get this working, I have my check book open

[dlokshin]

1. Not sure if the question is about how we technically do this, or how you go about doing this. The way you go about doing this is you connect your account to Dropbox (or Google Drive), drop the video in, and we take care of the rest.

 

2. Yup, we have a 9-axis IMU in the unit that we rely on heavily depending on the activity. Having raw accelerometer and gyro data is almost never useful, so it's a tricky proposition when making it accessible.

 

3. We had 3 Traces on John. Helmet Trace and tip-of-the-ski Trace performed very similarly. John did keep the tips out of the water most of the time.

 

4. Why stop at only a few million? :)

[MattP]

I can't wait to get my hands on one of these! @Horton‌ and I have been going back and forth with Trace conversations for at least a year. Great to see @dlokshin‌ dropping in to shed some light on the product. I know I would wear one every time I ski.

[Ed_Obermeier]

Pricing?

[dchristman]

Another interesting place to put one of these would be on the pylon in addition to the handle and the ski. Now we need a 3-pack!

[Horton]

Http://www.traceup.com

 

 

[AdamCord]

@dlokshin Thanks for getting back so quickly. With regards to the IMU:

 

How can we use this data in a meaningful way? I assume you've developed the software to use it for other sports. Even if we could see the Gs in each axis as well as pitch/roll/yaw side by side with the speed and position data it would tell us a lot. Hell I'd be happy if I could just dump that data into excel.

[Horton]

@AdamCord‌ imagine working on fin set ups with real data

[AdamCord]

@Horton I'm going to have to start taking my laptop to the lake with me. Fin setups, flex changes, boot changes, bevel changes...even Zero Off setting changes and different drivers will be quantifiable now. This will take the level of nerdiness in this sport to new heights!

[Horton]

@AdamCord‌ this from a man who has adjusted a fin with nickels and dimes

[oldjeep]

Too bad they haven't got a WP8 app, otherwise I'd have one on order to do some snow ski tracking.

[MattP]

@oldjeep‌ WP8... really...?

[oldjeep]

@MattP - yes, really. I know that it isn't quite the angry bird machine than an iPhone is, but for business use it is hard to beat.

[Skoot1123]

@dlokshin‌ - I am assuming you have two separate processors or receivers in the "Trace Puck"? One for the GPS and then one 3 axis accelerometer. Couldn't the sample rate of the accelerometer be different from the GPS? Granted if it (GPS) would update fast enough we would have the exact location of the skier. However, since you can already sync the GPS and video couldn't the accelerations be used to determine the vectors of the skier? The cross reference on this would be to correlate the vectors and the video - if the vectors state the the skier should be in "X" position and it corresponds correctly in the video then you know where you are positionally in the course.

 

Just thinking here: if the accelerations were sampled high enough we would be able to determine when the wake and turn are just by looking at the blips/changes. Knowing when the wake "happens" in the date trace from the g loading you should be able to interpolate between the turn and the wakes. Yes some sophisticated filtering would help take the noise out, but I am pretty sure it can be done.

 

@gator1 - you following this?

[dlokshin]

@Skoot1123‌ Lots of disparate parts here, and lots of GREAT questions, so I'm going to try and separate them:

 

* The GPS is the only chip that needs an antenna. So the IMUs don't have a "receiver" they just pump data out to the CPU as reqested. Each IMU has 3-axis, which makes for 9 axis total.

 

* In a perfect world, sensors would give clean reliable data without noise or drift. We live in a wold very far from this perfect state :). There's quite a bit to be careful of as well when using sensors.

 

- Sensors are only as good as their calibration. And calibrating sensors correctly is not trivial (we hold 3 patents on this alone). The second part is that different sensors have different strengths. For example: the Rooma (robot vacuum cleaner) has a gyroscope in it that's insanely accurate. It can vacuum your floor for an hour and then go, by itself, back to a charging base with almost no error. What an accurate sensor (I'm serious, here, a real feat of engineering)! Bring that same sensor into an action sport like water skiing and it's fairly useless (for reasons I won't go into).

 

- It's really exciting to use sensors (and trust us, we use them a lot) but a lot of times that's serious overkill. At 30mph you're traveling at 44 ft/sec. With GPS at 5 Hz, you're getting updated every 8.8 feet. It's hard to imagine that correcting with sensors would be better than some other approximation, for example.

 

- That being said, we have more than 10 TB of sensor and GPS data that we now use for our algorithm development. We've been working on this for a long time now. We have smoothing algorithms that assign confidence to a GPS coordinate based on the larger data set. So we ask ourselves "Is this possible?" or "Could there be something strange going on here?" When the answer to either is "Yes" we go back to the last point we have confidence in and then heavily rely on sensors to get us back on track.

 

- And then there are things that you can only do with sensors, and GPS won't help you: airtime, cutbacks, tricks, rotations, landings, etc.

 

[gator1]

@dlokshin‌ Thanks for explaining this all to us. To be clear, I am a strong supporter and hope your company can make this work.

 

At 50 mph, which most of us reach in the course, you are updating GPS once every 14.7 feet. A lot goes on in our sport in 14.7 feet, and we want(need?) to know most of it. !.5 feet means the difference between a middle of the pack pass and a record breaking pass.

 

Do you really believe you can fill in the 14.7 foot blanks with sensors and an algorithm? On top of the fact that there is error in the gps fix you got 14.7 feet ago?

[oldjeep]

@gator1‌ excellent point. That is one of the things that I am looking at in these devices. We hit 55 mph in a short downhill gs ski course. It is interesting to know the accelerometer data, but might be difficult to determine path at that speed.

[dlokshin]

@gator1‌ The accuracy you're looking for isn't achievable even if you take sensors into account (hand waving all the problems you're going to have there). What you're looking for is RTK GPS, which 1. Isn't really a consumer level product and 2. doesn't work well at high speeds.

 

The questions that should be answered: is this 1. worthwhile at all (maybe not, I just started to think about this sport) and 2. Is this considerably better than what you have now? Let me give you another example: John has two GPS beacons that help his boat stay at a consistent speed. There is some lag here, it's not perfect (GPS does have error in it especially if they use Kalman and not Doppler), but is it better than controlling manually? Absolutely.

 

Looking at John's data I think it's already pretty useful. And i didn't go into any backend systems to make it work. I just used Trace as is.

[Detrick]

This is great! This product could be a huge asset to our sport and become a component in our everyday training. Tracking path and speed allows us to dissect our technique even deeper than we all already do; hopefully giving us more answers on how to improve performance. I am excited to see @dlokshin‌ taking an interest in our sport, being involved and answering questions. As a skier and a coach, I see Trace products (once tweaked and finalized to adapt to our sport) being very beneficial if used frequently during training. I am looking forward to getting my hands on them in the near future. @dlokshin‌ I would be very interested to participate in testing your product to see how we can maximize the precision and for the use in our sport.

[GregHind]

@dlokshin that video of John was amazing. I'm glad you've taken the time to post here and share your expertise.

 

I can't fathom how your system works as well as it appears to if not using accelerometers in between GPS samples.

 

Perfect Pass boat speed control uses 5hz GPS to calibrate itself constantly and then relies on data from engine RPM signals and accelerometers to provide sufficient data for speed control

 

The Zero Off system in John's boat probably also uses some feedback about RPM (I don't know that for sure but it is highly integrated with the engine ECU and since PP needs the data and the RPM data is right there, I assume Zero uses it too). It also needs accelerometers to provide the accurate pull that it does. 5hz GPS is far too slow to be used for boat speed control alone.

 

Anyway, I'm not doubting at all but rather I'm very interested in what you've done, how you did it and what your system teaches us about our sport. Obviously there is some tricky, secret stuff in your device! Thanks for posting!

[Kelvin]

@dlokshin, Awesome product. I realize most of this discussion has been about slalom, but I'm wondering if the system could be used to calculate jump distances? If so, what level of accuracy do you think the system can achieve?