Distributed Human Action Recognition via Wearable Motion Sensor Networks Allen Y. Yang, Sameer Iyengar, Roozbeh Jafari, Philip Kuryloski, Ville-Pekka Seppa, Victor Shia, Posu Yan, Shankar Sastry, and Ruzena Bajcsy

© Copyright Notice: It is important that you read and understand the copyright of the following software packages as specified in the individual items. The copyright varies with each package due to its contributor(s). The packages should NOT be used for any commercial purposes without direct consent of their author(s).

This project is partially supported by NSF TRUST Center at UC Berkeley, ARO MURI W911NF-06-1-0076, Startup Funds from University of Texas at Dallas, Tampere University of Technology, and Telecom Italia Laboratory.

Project Roadmap This project seeks solutions to provide long-term monitoring of human motions and associated energy expenditure in normal living environments. A sensor network that supports such functionalities is called persistent. In order to support robust coverage of body sensor networks in both indoor and outdoor environments, we propose a scalable wireless communication system. The system consists of three components: Body sensor networks (BSNs) operated on each human subject. Local sensor networks (LSNs) that manage the communications between stationary wireless station and multiple mobile BSNs. Multiple LSNs interconnected via the Internet provide the wireless coverage for the purposes of persistent monitoring in both indoor and outdoor environments.
	Publications:
	Allen Yang, Roozbeh Jarafi, Philip Kuryloski, Sameer Iyengar, Shankar Sastry, and Ruzena Bajcsy, Distributed segmentation and classification of human actions using a wearable motion sensor network. Workshop on Human Communicative Behavior Analysis, CVPR 2008. [PDF]
	Level I: Body Sensor Platform
	The wearable motion sensor mote consists of three components. Top: A custom-built sensor board with a three-axis accelerometer and a two-axis gyroscope. Middle: A Li-ion battery. Bottom: A standard Tmote sensor network mote. The Li-ion battery is connected to a power module on the motion sensor board, and the motion sensor board is then connected to Tmote.
	Source Code:
	This technique is patent pending by the UC Berkeley IP offices. For licensing, please contact: [Office of Intellectual Property & Industry Research Alliances] Author: Allen Yang (c) UC Berkeley, 2008.
	Benchmark: Wearable Action Recognition Database (WARD) version 1.0
	We construct and maintain a benchmark database for human action recognition using a wearable motion sensor network, called WARD. The purpose of WARD is two-fold: 1. A public and relatively stable data set provides a platform for quantitative comparison of the existing algorithms for human action recognition using wearable motion sensors. 2. The database should steer the development of future innovative algorithms in the area of distributed pattern recognition by bringing together the investigators from the pattern recognition and sensor networks communities. WARD version 1.0 protocol: [download] WARD version 1.0: coming soon.
	MATLAB Sensor Interface
	We have designed a MATLAB GUI to sample, replay, and analyze the multi-sensor motion data. The software is free for academic users. Source code: http://www.eecs.berkeley.edu/~yang/software/WAR/SensorGUI.zip Authors: Ville-Pekka Seppä, Victor Shia, Posu Yan, and Allen Yang. Last Update: 6-10-2008.