Project Status: In progress. The game is expected to start in the middle of 2017
The idea of this experiment is about encouraging energy efficient behavior among NTU Campus dorm occupants for personal lights and ceiling fan. We build an online web portal that allows dorm occupants to submit their energy efficient actions and gain reward point based on them. Using cutting-edge IoT sensors we verify their overall performance, give them online feedback and potential incentivizing their behaviour to reduce energy intensity in residential halls in NTU Campus.
Design a stable and self-funded experiment, which will incentivize dorm/building occupants to reduce their overall energy consumption (ceiling lights/fans in dorm rooms). Moreover, apply optimization and machine learning techniques to model occupants as non-cooperative agents in a continuous game (either seeking Nash equilibrium or under a prospect theory utility representation).
This project integrate a verification method to proof positive contribution to the game. Occupants will be rewarded based on a lottery giving higher winning likelihood to those with larger amount of savings achieved.
The Eco Campus initiatives aim to reduce energy, water, and waste intensity in the Nanyang Technological University (NTU) – Yunnan Campus by 35% in 2020. In 2015, the floor area of undergraduate student residences has increased by 15% in comparison to 2011. The floor area of all residential buildings in the campus is projected to increase by 50% in the year 2020. As a consequence, energy consumed by residential buildings is projected to increase into 13% by 2017. Assuming all other sectors does not have increased in GFA.
This project aim to design a social game to reduce the energy intensity in the residential hall whereby the users will be rewarded with the savings achieved, and proof that a social game could improve energy consumption in dorm room.
Methodology and Stackelberg Theory
|Utility Learning Formulation|
Dorm occupants utility function under a game theoretic framework
agents decisions following approximate Nash strategy using inverse learning
Are the nominal and incentive components, respectively, of agent is utility function
Ms Geraldine Thoung