<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.28.3">
</HEAD>
<BODY>
<BR>
<BR>
Had a good time last night playing around with the Emotiv and Neurosky headsets.<BR>
<BR>
<BR>
With Rachel and Jake's help we tapped detections from the Emotiv headset into control commands for the MC Hawking electric wheelchair robot. I was kind of running back and forth between that and working on the RC Helicopters with Miloh but managed to get a couple second of video footage of the whole thing looking appropriately dangerous:<BR>
<BR>
<A HREF="http://www.youtube.com/watch?v=IPtuunvq8r8">http://www.youtube.com/watch?v=IPtuunvq8r8</A><BR>
<BR>
<BR>
Would love to get a more formal demo video shot, especially as we were just doing EMG detections (facial muscles) as opposed to "legitimate" thought control - although that's easy enough to do with the Emotiv. I just put that quick video together to send to NeuroSky to prompt them about the status of a pending SSVEP library. With SSVEP it should be possible to do precise 4-direction movements.<BR>
<BR>
SSVEP (Steady-State Visually-Evoked Potential) is very simple and requires no training. You basically tap into signals emitted by the the visual cortex at the back of the brain. You present a set of visual stimulus (such as checkerboard patters on a screen or even just a bank of LEDs) which blink a different - say 15 Hz for one, 18 Hz for another, and so on. The brain's visual cortex will harmonize with the blink rate in a detectable way, so you can tell which of a set of such stimuli the user is looking at. You then assign a direction control to each stimuli and you're all set - look at one pattern and you move forward, look at another and you turn right, and so on.<BR>
<BR>
If there's anyone here who's a strong coder with a solid mathematics background I would love to collaborate and could probably walk them through creating an Open Source implementation, provide the necessary papers, and put them in touch with the right researchers to support the effort. I think I can even dig up some code for MATLAB, though ideally the final version wouldn't have any proprietary dependencies. Please let me know if anyone knows anybody who might be interested!<BR>
<BR>
<BR>
Here's a great video of SSVEP in action from some guys I know at Northeastern University:<BR>
<BR>
<A HREF="http://www.youtube.com/watch?v=cuC0RTf1taw">http://www.youtube.com/watch?v=cuC0RTf1taw</A><BR>
<BR>
<BR>
I may swing by again this evening for a bit after the video game developer's meetup at Jillian's. Have run into a bit of a wall regards the RC Helicopters, about which I'll post a separate message to the list shortly - could certainly use some thoughts there.<BR>
<BR>
<BR>
Cheers<BR>
<BR>
Steve<BR>
<BR>
<BR>
<BR>
On Thu, 2010-11-11 at 05:14 -0800, Steve Castellotti wrote:
<BLOCKQUOTE TYPE=CITE>
<PRE>
Hey all--
I just wanted to send a quick word of introduction to the group.
My name is Steve. I swung over this past Tuesday evening along with
the Make:SF crew, and hung around the next several hours meeting a fair
few of you.
I've been managing an Open Source project for the past year which
is geared towards teaching kids (ages 10+ or so) a little bit of
neuroscience, helping them build robots out of LEGO Mindstorms, then
control and race them with their brains using consumer-grade EEG headsets.
Here is a brief demonstration video:
<A HREF="http://brainstorms.puzzlebox.info/index.php?entry=entry100923-100000">http://brainstorms.puzzlebox.info/index.php?entry=entry100923-100000</A>
And here is the project website:
<A HREF="http://brainstorms.puzzlebox.info">http://brainstorms.puzzlebox.info</A>
The current version of the software measures attention and
relaxation levels using a NeuroSky MindSet, translating those into
acceleration levels sent to the robots. Basic support for the Emotiv
EPOC is also available, although for classroom use I've been leaning
towards the former as it has dry sensors (where kids are concerned, wet
+ heads = bad) and is easy to put on and start using without much
fiddling about.
There's a variety of paradigms for controlling the robots and
several new types of "games" planned on the roadmap, but for the moment
the software is working and in active use in at least one classroom on
the East coast (Incidentally I am looking for more local schools which
might be interested to get involved). The focus is now on building up
case studies and fleshing out the teaching materials to better integrate
into existing curriculum.
Last night I brought round the remote control for a small RC
helicopter, and with a great deal of help from Milo, Anthony, John, and
a few others we managed to almost completely reverse-engineer the
circuit board and transmitter's communications protocol. I'd like to
extend a huge thanks to those guys for sticking around past 2 AM to help
bang it all out!
I'm planning to bring in my gear on Monday for the electronics
hacking session. I'll have the NeuroSky and Emotiv headsets, my LEGO
kit, and the RC helicopter (assuming I can managed to transport it all)
and would be happy to show anyone interested how it all hangs together.
With any luck I will already be on my way to getting the RC helicopter
to fly via the software. The intention is to pick up a second helicopter
and be able to have races in which two people compete to achieve and
maintain high enough levels of focus to keep the helicopters in the air
and be first to cross the finish line.
Thanks again for everyone's help and looking forward to catching up
with folks come Monday.
Cheers
Steve
</PRE>
</BLOCKQUOTE>
<BR>
<TABLE CELLSPACING="0" CELLPADDING="0" WIDTH="100%">
<TR>
<TD>
Steve Castellotti<BR>
Puzzlebox Limited
</TD>
</TR>
</TABLE>
</BODY>
</HTML>