SciPilot

Something I’ve been keeping an eye on lately is brain control interfaces i.e. devices which let you control things with your mind. There seems to be a sudden flurry of activity this year in this area with a small handful of products about to hit the consumer and research market! Ten years ago I tried to rig up a throat-mic and speech regognition software to save my poor aching hands, but it couldn’t cut it and besides, who want to be sitting in an office saying “OPEN FIREFOX”, “NAVIGATE TO FACEBOOK”. Mind control would be so much more efficient and private.

At the dumbed-down end of the market there’s a game called MindFlex which confuses some people with the levitation aspect of the game, but that’s done with fans. There’s only one sensor plus reference clip on the ear, so there can’t be many variables to control but the game looks quite fun. At least the actors in the promo video seemed to be geniunely enjoying themselves and had in fact briefly forgotten their failing acting careers.

At the other end of the spectrum is Honda’s research offering described in this great Gizmag article which combines two technologies – brainwaves (EEG) and infrared (NIRS) requiring the rather snappily dressed researcher to sit perfectly still while his slave Asimo is compelled to act out his intentions and wave its arms around. I’m not sure we’ll be seeing that in our living rooms soon, well, unless you happen to live at NASA or one of the darker military research organisations.

In the middle somewhere are two EEG headsets, the Neurosky and the Emotiv Epoc. My partner (who not co-incidentally does EEG for a living) couldn’t see how the Neurosky can be doing much more than detecting muscle activity as it only has one sensor located in a spot above the eyebrow most researchers avoid. However the Emotiv headset looks awesome and is aimed at both the game-developer and researcher markets with extensive SDKs and integration with the right scientific software. They claim to be able to detect three types of activity: actual active cognitive changes, the usual more indirect brain-wave levels and muscle activity to control avatars. It’s an interesting combined approach actually using the muscle signals instead of just being annoyed by them interfering with the true EEG. It’s the detection of intent that really excites me here though, as I can’t see using relaxation to help me do my day job. It turns out that the Emotiv was the brainchild of Dr Allan Snyder, amongst others, who I co-incidentally ran into once while briefly working for the Centre for the Mind at Sydney University. So it’s not suprising this piece of equipment is good, as he’s a very interesting chap.

So while I’m waiting for one of these headsets to appear, I decided to build my own partly for fun and partly for a science/art installation my partner and I are planning to demonstrate EEG technology (and for me to use Processing for something artitechnic). So I decided to write a new game to control as a test case – something we can project on a big screen and get kids to take turns in trying to play. We will also display raw EEG graphs and some processed signals to explain how the device works. So I convinced my partner to bring home some spare components from her lab but was slightly suprised to see the hospital-esque qualities of the syringes, cleaning swaps and contact gel… No wonder they have to distract the kids with a fake Hedwig before strapping the Davros-like helmet on them.

I thought I’d build the electronic interface with Arduino partly because I just want to play with it! So I’ve started (ahem, yes started…) planning the circuitry necessary to amplify the EEG signals without unintentionally re-programming myself at the same time. It seems I should be able to get everything I need to a couple of hundred dollars (AU).

So on to the game. I wanted a game which had a very simple control. I only have one sensor, so I’m assuming I only have one variable. This might not necessarily be true, as I could detect a range of brain-waves and variations, but lets keep things simple. The first game that came to mind was the old classic Lunar Lander! All you have to do is press one key to boost, and then let it fall as the lander moves sideways by itself. So what if the brain control is that, concentrate to fly higher, relax to fall.

I wanted to capture both girls and boys imaginations so I designed two versions – one with a been landing on a flower, or yukky bugs on either side, and of course the classic lunar version. With a little bit of help from Dreamstime and Processing viola! We have a little pair of games ready for the plugging in thereof the brain.

You can play the games here with the old fashion manual methods until I finish the sensor. Now where did I put that old 80′s headband…

The first spin off from the SciPilot launch pad was Motif – an online test interface for language impairment studies. This was the brainchild of Dr. Genevieve McArthur and Prof. Anne Castles from the school of MACCS at the Macquarie University, Sydney with a little help from myself to turn it into a web-application. I saw the web as a massive opportunity to distribute this free tool to a much larger audience, helping teachers and parents evaluate their kids and also potentially enabling MACCS run broader studies online. The previous tests were run on paper, PDFs and CDRoms with a mixture of issues, including borrowed laptops returning with kids’ pens jammed into them.

You can check out the site here at www.motif.org.au, but to see the tests, you have to pass some fairly rigorous approval processes! The site is intended for a serious research audience only. We are now planning to add more tests to the site, so it could become a serious tool in this area of research.

Technically speaking, I built the site with CodeIgniter, finally giving myself a chance to build a pure MVC application. I implemented the models with Deepend‘s CMS which provides an easy way to create  the administration functions, plus generating the models automatically. It was also my first experiment with Blueprint CSS to lay out the module grid which was very easy – after I’d spent a few weeks fighting with my own layout scheme. I hate browsers, I don’t have time for their quirks, give me a deterministic back end any day! I also used JQuery for most of the front-end interactive code, including the great form validation plugin – what a time saver! As I’ve often found JQuery combinations can lead to difficulties – I did have some trouble with using ReCaptcha and JQuery together, but it turned out to be the curvycorners plugin alone, so disabling corners in the user area was an honest repair. Oh, for -moz-border-radius in IE!

So at least I can say I finished one thing this year!

Welcome to SciPilot.

I promised myself 2009 was going to be the year of finishing things, so this site is partly dedicated to that cause. I have so many half-finished projects lying around that they’re starting to clutter my brain. It’s about time to begin to start finishing and get something out there at last. You never know, I might also slowly do a retrospective of my abandonded projects and ideas and just give them to the word. Open source them. Let the interweb do it’s tube-magic on them. There was the installation-art piece ridiculing the art-world value system, a purely XML-driven CMS with no backend, a quantum device which can cause propulsion with no ejecta, a UI framework driven by emotions, some awful poetry, terrible music and lots of dad jokes. So I’m probably going to struggle with the blog categories, but I’d sloppy kiss whoever invented tags.

Incidentally the name is a partial hat-tip to NosePilot, an old favourite of mine, a thing, uncategorisable really, but it makes the world a better place. If I could be remembered for something half as good I’d die happy.

So without further a do, there will now follow a pregnant pause … while I finish off the first of, hopefully many, possibly beautiful, things.