Energy-Efficient Human-centred Multimodal AI System for Sustainable Smart-home Living

The motivation for this can be illustrated through several situations in smart home environment. For instance, the system notices that the user is working in a room where multiple light bulbs are switched on, while another room has better natural light. Throughout, social interaction, the robot suggests: “The living room has more natural light now. Moving there could improve your comfort and reduce the energy for lighting.” On another example, the system may learn that the user usually wakes up later at weekends and spends Sunday morning in the living room. If the bedroom heating is left on, the system can turn it off, warm only the living room, and use natural daylight where possible. It may explain: “I have adjusted heating and lighting based on your weekend routine to save energy while keeping you comfortable.”

Beyond automation, the project will explore explainable and socially appropriate interaction strategies through human–computer and human–robot interaction. In this way, the designed system will not only make energy-efficient decisions, but also explain its recommendations, negotiate energy-saving, health-oriented, and wellbeing-oriented actions with users, and incorporate human feedback to continuously improve its decision-making process. For example, through observation, there are some vegetables in the fridge are close to expiry, it could suggest: “You have spinach and mushrooms that may expire soon. Would you like a simple healthy dinner idea using them?” This supports both food-waste reduction and healthier daily behaviour through personalised, context-aware recommendations.