Sensemble : a wireless inertial sensor system for the interactive dance and collective motion analysis

Abstract

The motivation for this project is the recent opportunity to leverage low-power, high-bandwidth RF devices and compact inertial sensors to create a wearable, wireless, motion analysis system meeting the demands of many points of measurement and high data rates. This thesis outlines the implementation of such a system intended for interactive dance, in which sensor nodes are worn on the wrists and ankles of dancers in an ensemble. Interactive dance is in some ways an ideal situation for pushing high performance requirements. Collecting data in a highly active environment of human motion demands a comfortable yet sturdy wearable design. Obtaining detailed information about the movement of the human body and the interaction of multiple human bodies demands many points of measurement and high resolution. Most importantly, using this information as a vehicle for interactive performance demands the real-time translation of data into an efficient feature set that a composer, designer, or choreographer can interpret. Now that it is possible to extend expressive motion sensing to multiple points on multiple dancers, an interactive system is capable of responding not only to individual motions, but also to how an ensemble is working together.

(cont.) The primary goal in this work is to demonstrate that simple features describing this type of collective activity can be extracted from the system and interpreted real-time, in order to generate responsive music or other immediate feedback. To this end, relevant strategies for feature extraction and music generation were implemented and tested, using data from a small dance ensemble. The results presented in this thesis show promising opportunities for future development in the areas of dance and interactive performance. In the broader scope, the hope is to expand this system to other applications, such as analyzing the dynamics of team sports, physical therapy, biomotion measurement and analysis, or personal physical training. Preliminary testing in these areas is also discussed.