Physical Sciences and Mathematics Commons^™

Lightlearn: Occupant Centered Lighting Controller Using Reinforcement Learning To Adapt Systems To Humans, June Young Park, Thomas Dougherty, Hagen Fritz, Zoltan Nagy

International Building Physics Conference 2018

Humans spend up to 90% of their time indoors, thus building systems should maintain the indoor environment within the comfort range. In this paper, we present LightLearn, a reinforcement learning based occupant centered lighting controller. The control agent interacts with the occupant non-intrusively, learns her/his preferences, and determines actions for achieving both human comfort and energy saving. We present system hardware, control algorithm, and experimental results of LightLearn for an office space. Compared to the full (9am-5pm) and occupancy based control, LightLearn reduced 83% and 63% of operation time, respectively, by adapting to the occupant.