Teaching Lights to Think

Lightlearn

A research team in The University of Texas at Austin’s Cockrell School of Engineering has an ambitious research agenda: shave even more waste from a building’s electricity budget by teaching lights to think. The results of their design and experiment, LightLearn, were recently published in Building and Environment.

“We would like to develop an intelligent system that understands the unique situation of an occupant and room environment and then decides the optimal action for both occupant comfort and energy savings,” says June Young Park, a doctoral student in the Department of Civil, Architectural and Environmental Engineering (CAEE). “Our research shows that you can save about 21% more energy without sacrificing occupant comfort by developing controller that adapts to each room’s conditions.”

Lighting consumes roughly 20% of the electricity budget in buildings. Historically, sensors have been used to reduce the amount of the waste because of lights left on longer than necessary. Passive Infra-red motion sensors, for example, detect motion in a room and switch lights on and off.

While motion sensor technology does save a lot of energy, the system has room for improvement. Two areas are the programmed time delay before lights are switched off lights and the unexpected cutting of lighting if a room’s occupant is particularly motionless.  Electricity is also be wasted if lights turn on while there is enough daylight. While luminosity sensors help minimized energy waste from this issue, they don’t account for individual preference in natural versus artificial light.

Park and his colleagues sought to increase energy savings by designing a lighting system that centers on the comfort level of a room’s occupant. LightLearn is a programmable system that uses reinforced learning to adapt to lighting preferences for each room’s occupant. The team tested LightLearn in five offices in the Ernest Cockrell Jr. Hall on campus.  The offices had different amounts of natural light, and occupants had varied comfort levels for lighting.

The team found that their system reduced the amount of time that lights were on when compared to traditional motion sensors, saving 21% in electricity. LightLearn was also favorably rated by the occupants.

“It is important to focus on the occupant because ultimately we build buildings for people,” says Zoltan Nagy, an assistant professor in CAEE. “Our goal is to provide a comfortable environment while eliminating energy waste.”

In addition to Park and Nagy, two UT Austin undergraduates, Thomas Dougherty (Mechanical Engineering) and Hagen Fritz (CAEE), worked on this research. Funding for this project was by both the Cockrell School of Engineering and Green Fund, a competitive grant program funded by UT Austin tuition fees to support sustainability-related projects and initiatives proposed by university students, faculty or staff.