Hacker News Reader

Comment by unwind | Hacker News Reader

Very impressive!

Not a word about of each "3d pixel" is implemented, which is a shame. I guess it doesn't have to be too complicated, but there sure are a lot of them even in this early prototype/lab version. They're also bi-directional, with each pixel having position-sensing.

View on HN

I've been working on something similar for several years. (Alas, I don't have MIT's PR budget.) I call it "PinThing" and implement the pins as DC motor powered, 3D printable linear actuators. I use limit switches to detect end-of-travel, and a shaft encoder to measure distance traveled. Current goal is to make an array of four 3x5 pin blocks to make a digital clock display. One of these days I really need to finish and formally publish this project! One big challenge in the project is driving down the cost per pin to make a large array affordable -- that's why I focus on using 3D printable parts and cheap motors.

Here's a demo of one pin powered by Arduino and an H-bridge motor driver circuit: https://youtu.be/pLzDHLm01wM

Three.js software demo: http://pinthing.com

well you can buy a chinese knock-off of the actuators the MIT people used from spark fun: https://www.sparkfun.com/products/10976

I'm trying to drive the cost down to $1-2 per pin. It's obvious the MIT team did not optimize for cost. At $20/pin (from the SparkFun link), a 28-pin display (the minimum you'd need for a 4 digit, 7-segment display clock would cost $560. To so anything interesting, you'd want dozens or hundreds of pins. Cost is a huge reason keeping this idea from being more wide-spread.

Well if you really want to drive down the cost, the best thing to do would be to mass produce the actuators for the purpose. The actuators MIT used aren't really intended as actuators, they're some weird part for moving knobs around on high end audio equipment.

Now as it turns out, there really are linear actuators that cost about $1 per actuator, they're just tiny: http://www.alibaba.com/product-detail/Goteck-1-5g-Super-Micr...

This picture shows what's under the table: http://tmg-trackr.media.mit.edu/publishedmedia/Projects/2013...

oh, they look pneumatic. I'd guessed solenoid. In retrospect a solenoid is either just on or off. pneumatics allow holding any position.

The prototypes I saw in person a few years ago were actually surplus automated potentiometers :) They were DC motors with gear reduction and a spool, with light steel wire coiled on the spool to slide the pot or the "3-d pixel" back and forth. these may very well be pneumatic.

A threaded-rod works, too.

True, although oftentimes way slower and getting the types of feedback they were using in the video would be tougher. On the upside, I'd expect this would work far better for heavier objects.

A regular threaded-rod would be slower, but you could speed it up by using a different thread profile with a longer pitch. If speed was the top goal, though, a belt-driven system could be as effective as pneumatics, but quieter. (But then you can't hold weight when the power is off.) Lots of trade-offs to consider, depending on design goals.

From their paper: http://tmg-trackr.media.mit.edu/publishedmedia/Papers/527-in...

it appears they use 900 motorized slide potentiometers connected to bowden cables.

Current prices from mouser suggest that the actuators alone cost $23,000. This isn't a cheap toy, but it does show what we could do if actuators got cheap...

Reply on news.ycombinator.com