Smart spaces are conceptualized as indoor environments responsive to user's presence and activities, and adaptive to user preferences and behaviour models. The emergence of solid-state light sources and Internet of Things are creating a range of opportunities in smart spaces to improve energy savings, promote human productivity, enhance VLC connectivity, and benefit health and safety. In the Smart Lighting Center, we are designing and developing the TiLED testbed to exploit applications and services of smart lighting technology.
Providing illuminance where and when needed by detecting presence and location of people and harvesting incident sunlight within lighting field; otherwise turn off or dim in time and space.
Providing task illumination by leveraging combinations of intensity and color temperature from luminaires to meet users' vision performances.
Visible Light Communications(VLC)
Projecting directional VLC to improve data connectivity by locating and tracking mobile devices' position and orientation.
Health and Safety
Enhancing sleep quality and general health with appropriate blue/cool light doses for human circadian rhythm. Integrating with in-home activity and emergency monitoring for the elderly.
The figures above show the TiLED Smart Spaces Testbed in PHO208,BU. Left picture shows the ongoing development with TiLED luminaire prototype, OptiTrack motion capture system and ubiquitous sensor networks. Right 3D model envisions future smart spaces that are fully programmable to deliver superb illumination, VLC data access, and a host of services.
This dissertation research develops a framework to enable a smart lighting room by optimizing lighting based on user activity information. With the assessment of different sensing modalities for suitability in supporting activity recognition, sensors are adapted to collect data about users and the room environment, and produce a rich set of features. An effective subset is selected and input to classification algorithms for user activity recognition. Lighting optimization is formulated with user activity information for objectives in energy, productivity and health. Thereby luminaires deliver optimal illuminance and VlC data for occupancy-sensitive lighting, task-oriented lighting, and beam direction for VLC. With appropriate sensors monitoring the actual output, this framework becomes closed-loop and capable of adaptive adjustments.
We are investigating prototypes for personal and clinical functional activity monitoring. A body area network solution involves the use of a PDA/smartphone and a set of wearable wireless sensors. The sensors, accelerometers and gyroscopes, capture 6DoF motion data. The smartphone aggregates and analyzes data in a human kinematic model. Multiple classifiers are applied to extracted features in time and frequency domain. An Extended Kalman filter is implemented for orientation tracking. Our approach can identify unconstrained activities in home and community, including postures, walking, biking, stair climbing; alert excess forward lean and risk of falling, and provide activity summaries in the energy expenditure and gait parameters.