Google Summer of Code 2008
http://code.google.com/soc/2008/openmoko/appinfo.html?csaid=55BC4E4D615BA291
Title: Accelerometer-based Gestures for Openmoko's Neo Freerunner
Student: Paul-Valentin Borza <paul@borza.ro>
Mentor: Daniel Willmann

Abstract: The emphasis is on showing how machine learning approaches can help 
recognizing an intuitive set of gestures that can easily be trained, and 
performed by the owner of a modern mobile phone. These gestural user interfaces were designed for accelerometer-ready mobile phones, like Openmoko's Neo Freerunner, using continuous density left-to-right hidden Markov models. This project 
touches on involuntary, and voluntary gestures, static, and dynamic 
acceleration, multivariate Gaussian classifiers used in endpoint detection, 
algorithmic implementations, runtime performance, and recognition accuracy. 
Sponsored by Google Summer of Code 2008 and Openmoko Inc., this project is currently continuously listening for a dozen motion gestures on the Neo Freerunner.

Videos:
Openmoko's Neo Freerunner does motion gestures, and screen orientation
	http://www.youtube.com/watch?v=K2S2rQUETwc
Accelerometer-based gesture recognition for Nintento's Wii Remote
	http://www.youtube.com/watch?v=MX50lXuJ28Q

Installation Instructions for Openmoko's Neo Freerunner
	http://wiki.openmoko.org/wiki/Accelerometer-based_Gestures

Thanks,
Paul