Integrating Leap Motion with Intel Edison | Hacker News Reader

Integrating Leap Motion with Intel Edison

medium.com

20 pointsvpj11 comments

To be perfectly honest, and from the perspective of someone who pre-ordered a Leap Motion on day one of their first demo video... This is about all the Leap Motion is useful for. It's actually pretty terrible.

I threw mine out couple of months ago totally convinced that noone can ever make it useful.

Sorry to hear that; have you had a chance to try it with VR?

This article should be titled: Integrating a device with no purpose with a Macbook and Intel Edison.

I was disappointed in how the leap motion was integrated with the Edison. The article uses a computer to talk directly to the leap motion and has a program send simple commands to the edison. You can do the same thing with an arduino board, no need for an Edison. I was at a hackathon where both the Edison and leap motion were available, but I couldn't figure out how to integrate the edison with the leap motion easily, or within a weekend. Wish the leap motion was a little more portable, especially for embedded systems like the Edison. That could really drive the leap motion's popularity.

One of the major problems with this is that whatever you're hooking the Leap Motion up to will need to be pretty beefy, since the way the LM works is by offloading the processing needed by the sensors [0]. The Leap Motion is in such a small form factor precisely because it doesn't actually do very much on the inside. All of the processing is done on the connected host. I have no idea what the minimum requirements for that processing is, but I don't think a RPi or Edison is going to cut it.

[0]: http://blog.leapmotion.com/understanding-latency-part-1/

The Edison has a dual core atom cpu. Don't really know how powerful it is, but it's pretty powerful for a small cpu.

It's pretty hefty, at least 1Ghz or so of desktop CPU. They've struggled to translate the leap to a raspberry pi/ARM platform because of the processing constraints.

Definitely -- being able to process 120+ frames of stereo image data, every second, with near-imperceptible latency, involves a decent chunk of processing power. That's why we recommend 2GHz at a minimum. The alternative would have been for us to sacrifice one of three things -- speed, accuracy, or low cost.

No. Nice try, but no, the latency is definitely perceptible. It really feels like Leap Motion's management went for only one of the "speed, accuracy, or low cost" you mention, and didn't really succeed at even that one goal.

Not to mention that it has no idea where your fingertips are during some gestures when it works at all. It's comfortable to use the LM for about four minutes, and then it's just a pain, combined with the enormous frustration of trying to get the damned thing to work right. Plus also something that really burned me personally: You guys really blew it with the deal with Best Buy. You turned pre-orders into "we ship them seven days before everybody can buy them at Best Buy". So I got my LM two days after I could have bought one at Best Buy. And supposedly pre-orders were ten dollars cheaper than buying one in the store (woo!), but guess what? Shipping costs completely nullified that.

I back a ton of hardware projects on KS and Indiegogo, and you know what term I use nowadays to denote a hardware project that's way behind schedule, and mismanaged, and barely works if and when you finally get it? I'll give you one guess.

I was super excited when the Leap Motion was announced, and I jumped at the preorder for one. That was however long ago it was, and it's been stored in it's box since. It's very cool technology but I have yet to find a problem that it's a workable solution for (sorry, Alex... I really want to like it.)

Reply on news.ycombinator.com