Home Automation: proposal for a low-cost and noninvasive approach

Smart home systems utilize multiple technologies and measuring devices to integrate into and control the typical home appliances. Additionally, they can automate and optimize energy consumption on interconnected systems, such as those controlling heating or lighting. In order for all of this to work, the system must learn about the habits of the house dwellers, incorporating this knowledge into their decision-making process. Presence detection typically involves complex and invasive techniques. This research describes a proposed design of a non-invasive approach using low-cost sensors and hardware, with the objective of gathering different perspectives of the environments in which the subjects reside. We present how captured data can be curated and shaped to deliver relevant value to the upper layers of the system, and evaluate different models employing artificial neural networks that could be used to make informed decisions affecting electrical devices attached to the system. This is an interesting and challenging approach which could ultimately provide features equivalent to currently available systems, but with the user's privacy as one of the most important drivers.