Reinforcement Learning based Smart Energy Management System for Indoor Lighting

The design of energy-efficient indoor lighting systems is a very interesting research challenge. Although this field has experienced significant developments, Reinforcement Learning (RL) remains relatively under explored. With reward-based optimization, RL seems to be the promising approach toward achieving sustainable and adaptive lighting solutions. The proposed approach introduces an RL-driven lighting control system that dynamically adjusts the brightness and orientation of LED lighting to optimize energy consumption while ensuring comfort for the users. The approach employs the Soft Actor-Critic (SAC) algorithm for processing of sensor data, related to light intensity, glare, and user response, to come up with lighting strategies at any point in time. The prototype system developed in simulation tests reveal that, compared to existing systems, energy usage can be reduced up to 25% without compromising the level of user comport. The results demonstrate the next generation of energy-efficient interior lighting solutions by providing a scalable, intelligent solution to smart building applications.