When & Where: 12:30pm, Friday, October 20, 2006, Room 200 AKW
The application of wireless sensor network technologies to everyday life situations still remains highly unexplored. To enable such applications, sensor networks in everyday life situations need to observe and interpret the behaviors of people in physical space and autonomously react to provide services. Furthermore, they are expected to do so with low latencies and without invading the privacy of their users. In this talk, I will describe a new sensor network architecture, the BehaviorScope, which aims to parse out human behaviors through sensor observations. A framework of probabilistic grammars used to transform low-level sensor data to high-level semantic interpretations. Our focus is to detect behaviors without tagging people with sensors. This talk will provide an overview of our approach and provide a description of the hardware platforms, operating systems and challenges we face in programming our system.