AI-Enabled Plug-In Smart Sensing Device for Household Load Monitoring

The Challenge

Residential electricity use is a major contributor to Australia’s carbon emissions and a growing source of cost-of-living pressure, yet most households—particularly renters—have no practical way to see how individual appliances drive their bills. According to the 2021 ABS Census, 30.6% of Australian households rent, many in apartments and multi-dwelling buildings where rewiring, switchboard access, or permanent monitoring installations are not permitted. Existing solutions rely on smart meters or circuit-level sensing that are often unavailable or impractical in these settings, systematically excluding renters from the energy intelligence needed to reduce consumption, change behaviour, and participate in emerging digital energy services. Without affordable, non-invasive, appliance-level visibility, a substantial share of the population faces rising energy costs with no means to respond.

The Solution

The project develops a non-invasive smart sensing plug that connects to a standard 240 V household outlet, enabling appliance-level monitoring without smart meters, switchboard access, or electrical modification. Two spatially separated plugs measure high-resolution voltage and frequency micro-variations; by jointly analysing these signals, the system achieves a form of electrical “stereo vision” that distinguishes genuine appliance activity from background supply effects. A novel foundation-model AI, developed by the Macquarie University AI Centre, interprets these signals using pre-trained, generalisable models that need minimal calibration, enabling rapid plug-and-play deployment. Households receive near–real-time insight into consumption, standby losses, and unusual usage through a mobile and web dashboard. Co-designed by Macquarie University and Smartizer and independently validated by the University of Technology Sydney, the technology advances from TRL 3 to TRL 5, with a clear pathway to commercial deployment.

Industry Partner