@Blake Thomas
See the Kickstarter FAQ here [ http://bit.ly/dvQyKD ] on why Kickstarter doesn't accept Paypal.

@Pablo González de Aledo
Yes, absolutely, the whole bit (design files, schematics, etc.). I just mean that my intention is that there will be an out-of-the-box structure I will target, but you'll be free to modify it however you want.

@Pablo González de Aledo

Fabrics meaning, the trained hierarchy of networks, structured into a standard form for other E1s to use. While my first priority is out-of-the-box experience, I have an eye toward modularity and would like long-term for people to be able to use E1 in the way you're thinking; as the core for projects I can't really even think of right now.

@Pablo González de Aledo

My goal is that E1's fabrics should at least be loadable from and exportable to some format or device that's easy for most users to work with (an SD card, initially), and that E1 designs can be easily shared.

Kickstarter's rules won't let you move the deadline after you've launched a project, but I will try again after the 6th and e-mail current backers. This is not the last chance for anything, it's just a way I can get it made. I will continue trying to raise money to that end, until they ship!

@Dacian Herbei

Basic synthetic intelligence is within the capability of embedded processors, but it matters a lot what you're planning to do and what kind of sensory processing you're doing. With an FPGA there's more flexibility; there's also much less expectation of a software stack being present so it keeps it very close to metal. The state-space of most modern robotics -- movement and some basic decisionmaking -- doesn't have enough "surface area" to pose a real challenge for the algorithms we're using. The obvious caveat applies: training matters a lot.

The problem for a copter is that a fresh E1 doesn't have any innate knowledge so it's easy to end up with a selective (read: destructive) training process. By default, a hexapod would be a little easier since when it runs into a wall, nothing happens, you can punish it, etc. When your quadcopter runs into the ground, not so much.

So this is one of the reasons for exporting the final state signals via header pins - teasing out some of those more complex patterns like hexapod kinematics can take a while, so it's easier to just look for high/low and write your own pattern generators, maybe on another board. This hybrid approach keeps overall logic usage low.

Hi guys,

I'll try to answer as many of these concerns as possible with the caveat that I just had a son less than 48 hours ago, so I'm actually typing this out from my wife's hospital room. For that reason I'm eager to get you answers, but I'm not really close to any hardware right this second.

First, yes, there's only a few days to go, and it's unlikely this will get funded to $30k in that time. (But if you guys send me an e-mail at josh@eksdyne.com I'd be glad to keep you up to date.) We've had a Kickstart up since the beginning of August, but a number of people have been confused by Kickstarter's all-or-nothing model and were unsure if they were buying something, or were being charged immediately, etc. And yes, it needs visuals for sure. Initially I was more embarrassed by the hackiness of the existing setup (see below) but hey, now I'll try and get some pictures once I get home. The presentation I mention below has a screenshot of a tool I coded up on OS X for editing some of the nets.

The $30k was also a lot high, but the problem was that to get kits down to a level where they were at $100-150, and that people would be getting kits and not just paying for completion of the project, I had to get quotes that for a much larger run than I would have if I had done a non-crowdsourced model. In other words, I could just put out a single board for everyone to look at for $5,570 (this wouldn't all be board cost, the thing itself wouldn't be that much) but nobody would get anything terribly fun out of it.

My prototype isn't a single board but rather a few Digilent boards, a modified Lynxmotion robotic arm with a 6.4" LCD on the end (think "Flight of the Navigator" here but less chrome, more epoxy), and a little bit an Altium Nanoboard I managed to pick up recently. It's a very fluid setup but the basic code is mostly the same across the different devices.. So E1 is a platform that needs glue more than anything. The tech on the inside isn't really complex at all -- I gave a presentation to a group in Blacksburg a while ago (http://slidesha.re/cXcZiF) that was a lot more technical, but the only thing I'm doing here is sticking it in one place.

@mpechner, I'd love to try and keep doing this stuff, but this Kickstart isn't in itself self-sustaining. I'd like to be able to make cool stuff, if there was a business around it, it would just mean I wouldn't have to Kickstart it. But I really have no interest in doing this stuff for "industry" or landing a big paycheck or anything like that. "Eksdyne Research", which hosts e1.eksdyne.com, exists only to accommodate the dorky stuff me and my wife pursue. I'm an engineer, she's a mathematician, baby.. I don't know what he does yet. Poops, I guess.

Anyway, please keep the comments/questions coming. I appreciate it. Thanks!

Josh