Ironically, the automation of building systems (a.k.a. smart building) still largely fails to satisfy its primary objective of providing human comfort due to the immaturity of the knowledge in Human Building Interaction. The recent development of personal comfort model suggested that the ‘one-size-fits-all’ approach of facility management is not feasible as human comfort is subjective by individuals and contexts. Therefore, the operation in the built environment should center human, further requiring adaptive solutions to learn different occupants efficiently. Therefore, the HBE lab used Reinforcement Learning to solve this complexity. Our prime examples include LightLearn, HVACLearn, and homieQ.
Researchers:
Steven Tanner McCullough, Nikhil Yaduvanshi, Terry Payne
Selected Publication:
- S. McCullough & J.Y. Park (2023), Occupant-centric system for healthy indoor environments via human building interaction, the 11th international conference on indoor air quality, ventilation & energy conservation in buildings (IAQVEC)
- J.Y. Park & Z. Nagy (2020), HVACLearn: A reinforcement learning based occupant-centric control for thermostat set-points, ACM Conference on Future Energy Systems (e-Energy AMLIES)
- J.Y. Park, T. Dougherty, H. Fritz & Z. Nagy (2019), LightLearn: An Adaptive and Occupant Centered Controller for Lighting based on Reinforcement Learning, Building and Environment