Competitive Agents for Intelligent Home Automation

Abstract

Technologies that aim to achieve intelligent automation in smart homes typically involve either trigger-action pairs or machine learning. These, however, are often complex to configure or hard to comprehend for the user. To maximize automation efficiency while keeping the configuration simple and the effects comprehensible, we thus explore an alternative agent-based approach. With the help of a survey, we put together a set of intelligent agents that act autonomously in the environment. Conflicts between behaviors, identified with a secondary study, are thereby resolved with a competitive combination of agents. We finally present the draft of a user interface that allows for individual configuration of all agents.