Friday, September 30, 2011

Experimenting with Kinect for education

We've had a Kinect at the Center for the last few weeks. Sadly, we've not been able to spend that much time with it (and is also not on the top of our priorities), but clearly see there are a lot of possibilities.

To me it seems that, in regards to educational purposes, Kinect will help out by engaging learners; forcing them to move just fires up some of those lazy neurons. By itself, this clearly brings an advantage to whatever app you hook it up to.

Imagine playing a math game with mouse clicks, vs. a Kinect-based game where you wave your arms, turn, crouch or jump to acheive goals. I think the latter might make reaching learning goals easier than the former. Of course, research has to be done =)

That aside, just getting the thing to work hasn't been that hard; basically there are several libraries out there that form the basic drivers on which developers can build on. The libraries seem pretty stable and work fine, but there aren't too many downloadable examples (or we're having a hard time finding them).

In future posts I plan to offer some updates documenting some of the stuff we've found. Some topics might be:
  • low-level SDKs or libraries that talk to Kinect
  • development stacks (from drivers to graphics/game framework) for Windows, Mac and/or Linux
  • ideas of real-world educational applications to build with Kinect (or showcasing existing ones)
For now, we've covered some of the basics:
  • Windows: installed the available OpenNI + NITE + SensorKinect binaries. On top of that, FAAST. This allowed us to emulate key presses through Kinect gestures. Thus, we can now have Kinect Tetris, Kinect Doom, etc. Interactive but not that innovative =)
  • Linux (Ubuntu): this was a tad harder. Some bits are available as downloadable binaries (and only then for Ubuntu 10.10), but most other require getting a bunch of libraries and then compiling. Not horrific, but not quite straightforward either.  End result: we're a step behind the work on Windows: just the skeleton tracking on some of the sample demos, no app control yet.
  • Mac: coming soon.
Why are we doing this?
I have to say this whole experience is interesting: I have almost zero experience developing native applications on any platform, which is our target audience exactly--students and teachers who want to build things, but need a kickstart (which we would like to provide).

Our goal isn't clearly defined at the moment, but we are confident this (like other technologies we work with) is a tool that will improve learning, or enable different ways of doing so. That might be an obvious statement to make, but part of our job is to actually test and prove it to be so.

And, I'd be lying if I didn't say I'm also having fun. Gladly, that's a part of my job too.

Wednesday, September 28, 2011

New OAI-PMH harvesting application: OAIConnect


Some months ago, I was hired to work on a web application that would grab metadata records from different document repositories (mainly in universities) and join them into a central catalog. This application should:
  • work with OAI-PMH repositories such as DSpace
  • provide faceted search of the harvested items
  • use open source products
  • be itself released as an open source product
The application was to be installed, and used by actual people in a project to harvest educational materials for a research project. That site is available at http://www.educonector.info/

Last week I released the code to the public as OAIConnect. The application can be installed by anyone who can meet some basic requirements (Apache, MySQL and PHP). The application is built on top of Drupal and a host of freely available modules (plus some custom glue), so you can basically extend, restyle and reconfigure it to your liking.

Screenshots
Installer:

Starting page:




Sample record: