ORIGINAL+ / Custom Skis Crafted With Artificial Intelligence

[mdf]

Good of you to make such thorough replies Sigi!

I'm sure I want a pair of the Originals. If I could only figure out what sort of skier I am, and where I prefer to ski ;-)

Although this was said in jest, it is the part I don't understand. It seems that most people are poor at articulating their own characteristics. I think I know what I like, but really all I can say is "I liked skis A and B, was indifferent to C, can't remember D, and hated E."

 

[tball]

Along the same lines, what are the inputs and feedback loop for the AI?

I scoffed at the AI aspect initially, but after a couple days of chewing on the idea, I'm thinking there could be some something there.

Any thoughts on patenting any of this? If so, maybe better to not discuss publically.

Thanks for coming here to discuss @sigiski! The Kickstarter price point is attractive. Hope you can keep it in that ballpark.

 

[sigiski]

Although this was said in jest, it is the part I don't understand. It seems that most people are poor at articulating their own characteristics. I think I know what I like, but really all I can say is "I liked skis A and B, was indifferent to C, can't remember D, and hated E."

CalG,

first of all, being able to articulate what you like and what not - we feel is completely sufficient to lead you to your "ideal" ski. It is not necessary that you can tell whether you need eg. more or less torsion, this is what we from an engineering & data view need to know.

Our experience from the test tour is that our concept is at first intriguing to better and expert skiers. We often faced a situation where a couple comes up our booth, and typically the guy (either skis better or thinks he skis better) comes get a test ski. When we ask his partner "do you want to test a pair as well", we quite often hear "ah, I'm not a good enough skier to notice a difference probably...". We still get those people on custom skis, and our overwhelming experience is that almost every level of skier will notice product differences. and will have a preferred ski he/she likes more than any other option. We really enjoyed this all winter, as it is a cool proof of concept.

We have a set of measures to understand WHAT people like, when the say the like something. The most important is quantification of subjective review statements through measuring and motion tracking (eg. a person describes himself as a "typically very fast skier" - we've built and developed a large enough data set that tells us what this precisely means (here the AI as a multifactorial analysis tool comes in extremely helpful).

Hope this explains it a bit better.
Best,

Siegfried

 

[sigiski]

Along the same lines, what are the inputs and feedback loop for the AI?

I scoffed at the AI aspect initially, but after a couple days of chewing on the idea, I'm thinking there could be some something there.

Any thoughts on patenting any of this? If so, maybe better to not discuss publically.

Thanks for coming here to discuss @sigiski! The Kickstarter price point is attractive. Hope you can keep it in that ballpark.

TBALL,

current input variables are:

1. Age
2. Gender
3. Weight
4. Height
5. Sole Lenght
6. Terrain preferences (piste, allmountain, sidecountry, etc....)
7. Preferred turn radii
8. self description of skill level with regards to speed, risk preferences, steepness preferences, aggressiveness preferences
9. max. of 3 ski areas that are your preferred ski locations

On top we currently also ask for the ski you are coming from/replacing.

As for patenting, you are very right, I cannot comment this topic publically.

Thanks, best

Siegfried

 

[tball]

Thanks @sigiski

Not sure, but some or all of that some of that ^^^ could be patentable as a business process in the US. Talk with an IP attorney if you haven't already. You have one year from that public disclosure to apply for a patent if you want to. It can be a provisional application, which isn't too expensive and buys you another year to improve the process and file the full patent application. You can then sell your skis as "patent pending" which could be part of your marketing and fundraising strategy. No need to respond, just want you to be thinking about it, as is sounds like you already have been.

How about the feedback loop for machine learning? What data are you going to cycle back into the algorithm? That will be an important piece to take the AI beyond just an expert system.

 

[sigiski]

Thanks @sigiski

Not sure, but some or all of that some of that ^^^ could be patentable as a business process in the US. Talk with an IP attorney if you haven't already. You have one year from that public disclosure to apply for a patent if you want to. It can be a provisional application, which isn't too expensive and buys you another year to improve the process and file the full patent application. You can then sell your skis as "patent pending" which could be part of your marketing and fundraising strategy. No need to respond, just want you to be thinking about it, as is sounds like you already have been.

How about the feedback loop for machine learning? What data are you going to cycle back into the algorithm? That will be an important piece to take the AI beyond just an expert system.

Hello tball,

thanks for your thoughtful input - really appreciate that! If some of the following is not 100% perfect english, pls excuse that as this now goes beyond my daily english skills...

one important fact with regards to looping back is the fact that every ski that leaves production is measured in variable dimensions like flex (pattern), several torsion values, etc. all of that building a large database of individual (even if you might call them semi-customized as some dimensions as sidecut are fixed) skis; along any future process or feedback loop, we refer back to this individual ski data

with regards to the feedback loop, I have to limit my answer somewhat here, as you are right with your inputs or questions on IP protection. it seems you are well into AI, so I can say that much that our core application is a CNN; we have three basic training loops for the system, one of them being our internal test team procedures, the second being our ski test tour and the third being after-sales feedback we invite purchasers of O+ skis to give it to us. for anything further, I'm happy to take this off-public, for said reasons. feel free to contact me via https://www.linkedin.com/in/sigi-rumpfhuber-4693b3/

thanks!

 

[cantunamunch]

Ski pair #81 has arrived and ...they are pretty.

 

[Erik Timmerman]

Wow, you got a pair. They do look pretty sweet. Are they for you or herself?

 

[cantunamunch]

She doesn't want any more skis - guess I'll have to ski them

 

[dbostedo]

Put demo bindings on them and bring them to whitetail or roundtop!

 

[cantunamunch]

Put demo bindings on them and bring them to whitetail or roundtop!

Bwahahaha.

I did back away from my original plan, which was to use Piston plates. This pair doesn't need them. They are damp, yes, but not in a Head Monster way. They don't thunk. They hum! It's almost like O+ tuned the fall off to almost a single tone, for feedback to the skier, or for fun. Surprised me, honestly, and not a little bit. It's like hearing a heavyweight boxer hum the 'Zip' of Zip a dee doo dah when he takes a hit.

Sorry, chief. I'd love to let you try them but putting a demo binding on a ski with this sort of torsional flex is ...not going to happen. Bring your 305mm race soles.

 

[oldschoolskier]

I’m going to take a pretty good guess that AI is trying replicate consistently the process of the Ultra Talented Ski Artist Master Constructor (with a little luck in organic material selection) is doing to produce those ultra WC winning master piece skis.

The AI quantifies the “feel” portion of the Master while eliminating a lot of other subjective variables in organic raw materials.

Simple concept, little more difficult in application. Will be interesting to watch as this progresses and the potential impact it has on the industry.

Final note, the only limitation with AI is the intuitive jumps and unconnected advances made by humans still can’t be beat. Replicated after the fact maybe, but not before.

 

[Philpug]

I am really intrigued by these. The story behind them is pretty cool. Good going @sigiski.

 

[dbostedo]

...I'd love to let you try them but putting a demo binding on a ski with this sort of torsional flex is ...not going to happen. Bring your 305mm race soles.

Boo... you're no fun! Don't you know all the cool kids are using demo bindings these days? (And just ignore the Marker X-cells I put on my Stockli's... )

But maybe I can still see them out on the hill sometime this season.

 

[cantunamunch]

But maybe I can still see them out on the hill sometime this season.

No need to wait - bring a nice bottle of wine over and I'll let you put prep wax on them? Oh, wait, you don't wax

They are a Bad Kleinkirchheim/Deer Valley/Tremblant (and pretty much in that order of priority) build; Whitetail not so much. The flex is amazingly progressive - two fingers to squeeze out the camber but then you're in OMG/ludicrous mode.

 

[skibob]

Oh, wait, you don't wax

Why attack his personal grooming habits? Sheesh!

 

[cantunamunch]

Why attack his personal grooming habits? Sheesh!

I'd be a lot more relaxed if he was going on a hypothetical date with my hypothetical daughter than going near these with a hot iron, yes it's true.

 

[James]

Ski pair #81 has arrived and ...they are pretty.
View attachment 56924
View attachment 56925

View attachment 56927

Wow, nice job purchasing these! Geez, easier to get than Crocs.

Can you tell us how you ended up with your design? What general type of ski did you want to end up with?

 

[cantunamunch]

Wow, nice job purchasing these! Geez, easier to get than Crocs.

I'll assume you're referring to Augment née Croc, not the rubbery footgear

Can you tell us how you ended up with your design? What general type of ski did you want to end up with?

Not much to it: I wanted an all-around mixed use with short turn bias, skiable South Alps/NE/Colorado in that order. I gave them my physical stats, told them I liked the Monster 88 and old VXL, told them I wanted faster rebound than the M88, and told them I was in a more upright boot for this season.

I didn't get the Tyrolia system binding because I thought I would be using Piston plates. As discussed above, I realised that would be inappropriate. I went with STH instead of Mojo 15, just because I wanted to.

 

[tball]

Not much to it: I wanted an all-around mixed use with short turn bias, skiable South Alps/NE/Colorado in that order. I gave them my physical stats, told them I liked the Monster 88 and old VXL, told them I wanted faster rebound than the M88, and told them I was in a more upright boot for this season.

Do you know if the AI customized your skis or one of those pesky humans?

How about the length? You probably told them the length of you Monster 88's?

Very pretty skis!