Passive Heating-Cooling of Buildings using Reinforcement Learning

The traditional practice to do air conditioning in a building is to isolate it from the environment for ease of temperature control. We challenge that notion to argue that using climatic resources effectively and intelligently handling the doors and windows with careful control mechanism, we can minimize energy consumption significantly. Our goal is to find that path using Reinforcement learning and proposing a generalized agent that will implement transfer learning to use its knowledge to provide optimal performance in any building, in any weather.