This project by Tim Sherratt caught our eye for just these reasons and clearly demonstrates this methodology most clearly.

Some exhibition magic with Zotero and Omeka

We’ve looked at both Zotero and Omeka in the past for various projects because of their growing use by the research community. This project leverages existing data with an API, an open source CMS and some Python magic to create a new view in to an ongoing research project.

“Thunderbolt is a temporary public artwork created for the 10th anniversary celebrations of the Sydney 2000 Olympic Games.

“Made from a recycled windmill found in Broken Hill, the nocturnal lighting of this dynamic meta sculpture changes in repsonse to the amount of electricity consumed in the nearby neighbourhood.”

Thunderbolt is a public sculpture piece by Bonita Ely of the The Environmental Research Initiative for Art at CoFA UNSW. Version 1 is installed at Sydney Olympic Park and will run initially for 3 months. The lighting of the structure is powered by solar cells and altered based on the power consumption of the surrounding residents. Long time collaborators Allan Giddy and Rob Largent have put together the lighting control rig, EnergyAustralia have kindly donated their data, and Laudanum have provided the glue between the two.

On site we have a tiny Microclient SBC connected to an Arduino board and a 3G modem. Each of these 3 parts are seriously overspec’d for the the job but are all part of our usual toolchain and were on hand. The lighting rig expects its 5 inputs to be tickled to which it responds with a mix of its 3 lighting colours. So our computer connects to its data source and tells the Arduino to bring up one or more of its pin outs.

In order to work around a couple of techno-political issues (the temporary nature of mobile broadband connections, and who connects to whom, when and how) we elected to move the central data source to Google App Engine. Having done that both the data provider (the company providing our stats) and data consumer (the computer driving the lights) were able to abdicate all the data munging to the App Engine too (which was a boon).

Some GAE gotchas.

• Large XML archives of prior data to pre-populate our database. Fine consumed in small chunks but quickly exceeded the GAE quota
• Unique keys are provided for each `put` in GQL – but you can specify your own key thereby making a UNIQUE KEY field. We hashed together our timestamp and station ID to use as a key. Each time GQL puts with the same key it REPLACES the row. There is also a get_or_insert GQL method which either gets existing data for a keyname or puts new data where the keyname doesn’t exist.

Parts

• Microclient SR
• Huwei mobile broadband USB modem
• Arduino Duemilanove
• Ubuntu, Python, Google AppEngine

Set up our build environment and get the necessary packages. You’ll also need your gpg key ID if you’re going to sign the package.
$ sudo aptitude install bison flex libbluetooth-dev libgtk2.0-dev python-dev libtool automake1.10 autoconf quilt patchutils python-all-dev cdbs
$ mkdir cwiid
$ cd cwiid


Get the source, unpack and patch.
$ apt-get source python-cwiid
$ tar zxf cwiid_0.6.00+svn201.orig.tar.gz
$ gunzip cwiid_0.6.00+svn201-2.diff.gz
$ mv cwiid_0.6.00+svn201-2.diff cwiid-0.6.00+svn201/
$ cd cwiid-0.6.00+svn201/
$ patch < cwiid_0.6.00+svn201-2.diff


Get the patch we want to apply (from http://abstrakraft.org/cwiid/ticket/95 & http://abstrakraft.org/cwiid/ticket/89)
$ cd libcwiid/
$ wget http://abstrakraft.org/cwiid/raw-attachment/ticket/89/0001-added-low_speed-data-from-motionplus.patch
$ patch < 0001-added-low_speed-data-from-motionplus.patch


Configure
$ cd ..
$ ./configure


Build the package. The deb files are placed in the directory above.
$ dpkg-buildpackage -rfakeroot -b -k474f6948
$ ls -l ../*.deb


Install the packages
$ sudo dpkg -i ../*.deb


Install libusb from MacPorts

whale:pyusb-0.4.2 rossetti$ sudo port install libusb

Patch the setup.py file for pyusb-0.4.2

37a38
>     '-L/opt/local/lib',
39c40
<     extra_compile_args = ['-I/sw/include','-I/usr/local/lib']
---
>     extra_compile_args = ['-I/sw/include','-I/usr/local/lib','-I/opt/local/lib']

After that pyusb should install well.

whale:pyusb-0.4.2 rossetti$ sudo python setup.py install

Wade pulled together a small team to build a prototype semi-autonomous robot in a Victorian dress. Our job was to provide an interface between the web and the controller board. We installed a small webserver on an EeePC netbook. An Adobe Flash socket sent mouse clicks and speech instructions to a tiny Python server that interpreted whether the instructions should be forwarded to the controller board or handled by the EeePC. Speech commands were forwarded to the onboard Festival server using an MBROLA english male voice and then piped to the sound card.

We used small form MicroClient SNR computers running Debian and some funky Python scripts. The system was quite polite – should you refuse any message it would leave you alone henceforth.

The Stride was the artwork we tested with. An installation depicting the final 100m of Cathy Freeman’s gold medal 400m sprint. We sent the public a text description, then a video of the race, finally an image of the site. Response was much better than expected given the public’s general suspicion of Bluetooth in general.

We took an Eee PC connected to an old Garmin III running Kismet. Lid down. Earpiece pinging. Raining. Shoes with holes. I’m certainly glad the police didn’t tap me on the shoulder. Go rider!

We post-processed the logs in Python to generate Google Earth KML files depicting the coverage. For an old girl the Rocks is certainly rocking to the WiFi.