computationalhumanist.org

Technological archeology – resurrecting the Qwerk

HumanUnit0 — Thu, 30 Sep 2010 02:53:03 +0000

The Qwerk is a robotic controller made by Charmed Labs, a spinoff of Carnegie Mellon.

It can control beaucoup motors and sensors and has USB, Ethernet, and serial ports for communicating with just about anything. It is cheap and durable. It can be run over the Internet by the Telepresence Robot Kit, TeRK, which is another Carnegie Mellon project.

It is a little Linux computer with a big grasp. I bought three, and now it is no longer being supported.

Actually, VEX has bought it and is spiffing it up to be the star controller of its new line of robotic parts, VEX PRO.

But the VEX PRO controller hasn’t come out yet, and I still have three Qwerks –– and a need to use them.

So we are going to do some technological archeology.

Our resident Linux wizard is focussed on getting the ARM gcc keychain built on Ubuntu 10. Everyone else will concentrate on programming the Qwerk. The Qwerk developers were committed to learning by doing. Which is my polite way of saying that the Qwerk is poorly documented – its developers believe that you should just meditate on sample code and the Zen of Qwerk will be with you. Or as Andrew Kirillov puts it:

If you open a Qwerk box, you will find there quite short note with information on how to start. Basically it says: “Visit Qwerk Start page for information on how to get started”. That is all. And the Qwerk board in the box of course …

As if things weren’t obscure enough, the Qwerk is programmed in C++ but TeRK runs on Java. Fortunately, to run the Qwerk autonomously, we only need to use C++ with libqwerk. We will begin by mastering the Qwerk motor commands in libqwerk and try to pick up the C++ that we need along the way. I have put the files for the flower project on the PBworks wiki, as well as the source files and their text version for us to comment.

I will gradually add the relevant motor control files from libqwerk for a similar exercise in code exegesis and subsequent commenting.

Three high signal-to-noise cheers for Cedric Walker!

HumanUnit0 — Wed, 29 Sep 2010 16:38:10 +0000

One of our projects is to build a sparse-array EEG system as the brain part of a brain-computer interface (BCI). This was originally the brainwave of my student Alan, who wanted a BCI for producing music from the electrical activity of the brain. (Alan also got me interested in piracetam, but that is a different post). After rummaging around on the Internet, we came across the OpenEEG Project, which was originally conceptualized as a cheap way of introducing people to biofeedback. Nowadays, it has become a conduit for labs and researchers like us who are trying to build BCIs on the cheap.

The hardware part of the project consists of a two-channel EEG system composed of an analog circuit board for collecting the signal and a digital board for digitizing it and sending it out over a serial port. The boards are built by a Bulgarian electronics company, Olimex, for about $80 a piece. Alan got a grant, and we bought a bunch of boards plus enough ancillary equipment for a single two-channel system. My students put it all together but ––– it did not work. After much fiddling and frustration, it still did not work. This was a large disappointment, because it absorbed much of our attention last semester.

A few weeks ago I met Paul Tran, who nurses computers for Psychology. Paul helped me extract some data from a Jurassic PC. It turns out that he is an electrical engineer, but for systems analysis, not circuit board diagnosis. But my juices were flowing, and I got in touch with the Electronics Lab run by the Coordinated Instrumentation Facility. But they lost their electronics guy after Katrina and have not been able to hire a new one, though they thought that Biomedical Engineering might have such people. I asked my contact in the SSE Dean’s office, who put me in touch with Cedric Walker, who I had met at a robotics gathering at the New Orleans Charter Science and Math High School.

I took my zombie EEG box to Cedric’s office Monday morning and got a fascinating lesson in circuit-board troubleshooting. He immediately determined that there was an overload, presumably from a short circuit (the LED did not come on, and the power supply got hot to the touch almost immediately). He methodically checked through the obvious things that could go wrong, mostly involving the cables that we had made and connected. He finally concluded that the XLR connectors were not connected correctly. He gave me lots of tips, like how to use the sound card in a PC as an oscilloscope. He also had a thermal wire stripper which was too cool for words. Stripping the thin wires that we worked with incorrectly is about the easiest way to screw up. So I immediately ordered one.

When it gets here, I will get started on repairing our poor beastie.

Back to blogging — networking the motherblog

HumanUnit0 — Tue, 28 Sep 2010 17:30:30 +0000

By chance I sat in on Gardner Campbell‘s keynote address at Tulane’s Tech Day last Friday and was fascinated by several things he said, and showed. He started out talking about what makes something interesting, drawing on Paul Silvia’s book Exploring the Psychology of Interest (2006), which appears to summarize a thread of psychological research going back several decades. Silvia discusses four factors that create interestingness: novelty, complexity, uncertainty, and conflict. Campbell goes on to apply them to new media in teaching. His first example was BlackBoard, which is ––– (what’s your guess?) –– not interesting. He contrasts it with the “motherblog” for the course that he is teaching this semester at Baylor, the New Media Faculty Development Seminar, which certainly looked interesting to me.

Then I had to leave to get on with my life.

After much digging, I found out that Campbell’s motherblog is a WordPress network of blogs. ComputationalHumanist is a WordPress single user blog. So now I am debating whether it is worth the effort to turn it into a networked blog for my students. It’s not much of a debate, however, because I have already asked them to start blogging as a way of documenting their research, which is a way of advocating blogging as the new media version of the laboratory notebook.

NB. This is my first post since I set this blog up three years ago. I suppose that Campbell should be pleased that his presentation stimulated such a response.