Rubick's cube - LED, networked, automated

[Dave3]

I have been trying to solve the rubick's cube quickly. So I thought it might make sense to make a automated cube, replacing the colored stickers with LEDs. Some advantages:

- Instant scrambling of cube by pressing a button

- Could practice specific solves over and over again by resetting LED positions via the press of a button

- Cube could follow all moves and timing, and suggest different moves or identify areas to improve

- All important cool factor

Someone tried to launch a kickstarter campaign with something somewhat similar. He/She had a cool idea for international competitions via the internet. But I think his design is too complex with a ton of parts (27 custom made batteries per cube!?!). Campaign shut down.

Still a great thought expirment. It would also be very useful for a few hundred people around the world trying to break solve records. How would you try to improve design?

https://www.reddit.com/r/Cubers/comments/49pwwb/new_led_lit_smart_speed_cube_campaign_launched/

Some preliminary thoughts:

1. The cube has 6 center cubes that are connected by a center spindle; importantly, they are always in the same position

2. I think the smart cube would need 6 high speed / high precision rotary encoders to track those "6 center cubes"
 
 a. maybe the positions of all the other cubes can be calculated with just those 6 encoders - still thinking about that
 
 b. expert solvers can do 3+ abrupt turns per second so that might be a challenge for small encoders to follow accurately

3. Also need to provide battery power, battery wires, color commanding wires to all 64 LEDs without wiring getting twisted up

 a. Also will need color controlling system.

4. Cube will need some internal processing chip for the resets and maybe to save a few positions. Alternatively, that could be done via bluetooth to an external app

5. Transparent plastic cubes are not too durable and will leak light. So probably use a black plastic cube off the shelf, drill a hole in each cube side to insert an rgb LED.

6. Not much space inside cube. Need to keep weight down. Keep balanced.

If you guys come up with some viable ideas, I would be happy to give them a stab and document here. This is just for fun as this is a hyper niche market...

[PlainName]

cube would need 6 high speed / high precision rotary encoders

Just contacts which the outer cubes make. If the outer cubes had info encoded in some way (resistance, perhaps, or I2C) the bridged contacts tell what cube is there and how it's oriented. The drive for the LEDs can be via contacts as well, or if you wanted to get really clever the outer cubes could be simple light pipes.

I think the way this might be possible is by making all the outer stuff pretty dumb. Twenty-seven batteries is a bit of a joke - even if you went for 27 smart cubes, the power could be supplied from a single battery.

[Dave3]

Light pipes to the outer cubes sounds like a great idea!

The 9 corner cubes and 12 edge cubes are always shuffling around so maybe there is a center of 54 RGB LEDs which remain in one place (along with the 6 center cubes). But when the user twists a side, the LEDs turn off very briefly for that side. The LEDs are reassigned by a controller, then the LEDs turn on showing the new pattern...

A few factors might challenge electrical contacts or optical readers:

1. Good solvers move the cube extremely fast and violently.

2. Each user tends to use some sort of silicone / water / unknown lubrication which could mess up any electrical contacts or optical readers to the cubes. Interestingly, some lubrication is to SLOW DOWN the cube

3. The center cubes are connected to the center via a point plastic axis. Each cube has a screw and a spring which are used to adjust "tension." Some users prefer very loose, flexible and sloppy cubes which would especially challenge electrical contacts.

I like your idea of assigning a different resistance to each cube. Then using , maybe some type of springed electrical contacts (thinking the brushes on an older electrical drill motor) could work. The contact only needs to be instantaneous and not for 100% of the cubes as some logic could be constantly scanning the cubes and with some % of resistance data (<100% I think) calculate the position of each cube; then assign LED colors appropriately...

To your

[PlainName]

some sort of silicone / water / unknown lubrication

That could put a spanner in the contacts, yes. Although I would question whether this kind of thing is appropriate to that kind of user - the only gains over a standard Rubik are instant shuffle (useful) and prettiness (not essential).

By contrast, for a beginner or player of medium skill, the LEDs could be used to hint at solutions, so there is a practical  benefit to them other than just being pretty. And that kind of user isn't going to working at the speeds that would be a problem, I think.

[Dave3]

An engineering buddy suggested that wiring the LEDs and related controls is not pratical.

1. One option is a battery + wireless communication module but he thinks that would be very expensive. And probably too bulky and power thirsty with the modules he knows.

2. A simpler option he suggested is to use a cube with no color stickers as a physical control. But a virtual cube with correct colors would be viewed from a computer screen. This might be done with 6 magnetic rotation sensors inside the cube to record the cube twists. Then a small bluetooth transmitter would send rotation data for the six channels. There a computer could calculate position of each cube and display a virtual cube on the screen.

I think this later option would also require some sort of position detector so the screen cube is positioned correctly with respect to the physical cube.

[Dave3]

Thanks for the ideas here guys!

OK-  massive simplication to focus on core functions of the cube vs. elegance. The core functions of the cube need to be:

1. Ability to reset LEDs to "solved"
2. Ability to set LEDs to a specific scramble on the computer (so that one can practice a particular solving algorythm)

The simplified system will have three main parts:

1. Cube
- 54 Assignable RGB LEDs (WS2812B?)
- 26-54 rechargable batteries / supercaps
- 3 pads per LED (+, -, data)

2. Programming Box
- Put cube into box after each solve to reset LED pattern & recharge batteries

3. Computer

Each of the 26 individual "cubies" is very small, and each of the corner "cubies" will need 3 LEDs. So we need the smallest LEDs & batteries possible. Supercaps might be usable if they are small enough and can last say 5 minutes at low intensity before recharge. Any recommendations?

[bookaboo]

Why make it physically rotate? Make a single solid cube with some sort of moving switch, or a capacitive touch with haptic feedback along the verticies or corners. The design challenge becomes one of making a convinving and intuitve user interface...... but I'd rather do that than deal with mechanical issues.

[Dave3]

Thanks for the feedback bookaboo.

Some of these exist in physical form but they are not great at locking in "muscle memory" and reflex.

http://www.futurocube.com/

http://gizmodo.com/5066330/led-rubiks-cube-is-unnecessary-digitization-of-a-classic