A special glove that turn all 3x3 into smart cubes!

[Wish Lin]

Please read with patience↓

If you don't know, I have been on a quest of revolutionizing smart cube technologies in hope to make them cheap and accessible to the majority of the cubing community.
Here are the attempts:

But, they all turned out to be a bust for different reasons. I don't think I can come up with any new externally-placed-sensor smartcube ideas.

BUT.

A totally different new idea came into my mind. After a long discussion with @Lelouch , a forum member who has experience in neural networks(i.e. machine learning), I am excited to show our progress so far!

Rough Draft:

It uses 5 flex sensors positioned on the fingers to capture the user's finger movements(flexing angles). These data will be sent back to a phone or a computer via bluetooth for further processing. Here is where the magic happens: a pre-trained neural network will analyze these data and determine the specific fingertricks being performed, and thus achieving the goal of anycube becoming a smartcube!

Some more progress has already been made, but I want to release them to you guys later after both of us had an agreement. If you are interested in this project, please post your thoughts below!

 

[ProStar]

I'm really hoping that this becomes a reality. But based on your track record, I can't help but question whether this actually is going to happen

 

[Wish Lin]

I'm really hoping that this becomes a reality. But based on your track record, I can't help but question whether this actually is going to happen

Yes, every time I come up with a new idea, it always looks so promising. But when I actually conducted it always has something I didn’t expect that prevents it from happening. And of course I cannot guarantee If this will be the same or not.

But so far, I have already buy more than $150 worth of material and over 30 hours on the CAD and programming on this, so I AM trying my best, as a senior high school student, to make this become a reality.

And even if I failed(there must be some bottleneck ahead), someone in the future can still use these data and maybe actually make it, this project or the previous ones.

 

[aisojdb]

cool

 

[qwr]

machine learning is not what is going to make this project happen. it's understanding how to build reliable electronics and hardware devices which is a totally different area. I have serious doubts ML would even be better than traditional DSP approaches. source: I did my share of hardware hackathons and in grad program for ML.
what concerns me is that the rough draft only accounts for finger movement and not wrist movement which is just as important.