We worked with Sense during the prototyping phase of their IoT home energy tracking device. Their vision was of a home energy monitor capable of identifying different electronic devices in the home and giving homeowners insight into their patterns of energy consumption, alerting them to trends as well as ‘energy hogs’ and inefficiencies. When we joined them, they were still perfecting their machine learning models and needed assistance in visualizing the raw inputs to these models. We selected, designed, and implemented the entire technology stack for the project, from Front End to Database to cloud-based software platform.
Devices such as the Sense Home Energy Monitor have their own specific set of data challenges that center on processing, analyzing and, above all, visualizing results from such a large and constantly-flowing data stream. With energy use data being collected on the sub-second level for periods of days and weeks, Sense needed a system that could ingest, filter and analyze massive amounts of data as well as present it to their customers via a simple and intuitive visualization. We needed to design a system that could rapidly load data and function with minimal administration-something that could largely run on autopilot.
The ability to display household energy consumption in an easy-to-read manner is one of the central selling points of the Sense monitor. It would not be an overstatement to say that the entire project hinged upon a flawless visualization that would function without delays or hang-ups. In order to be visualized, the continuous data had to be smoothed to account for discreteness, and the visualization itself had to have the capacity to display multiple time series and support zooming and panning. Such functions would allow customers to focus on specific time periods or devices as well as the big picture of their energy use. To accommodate these functions we turned to the JS libraries typically used for such visualizations, but none of them did exactly what we needed. In the end we had to modify an existing library to get the job done.
We constructed the entire data infrastructure for Sense, bringing the final product from vision to reality. The software platform that we developed runs on the cloud, and the system gives Sense the ability to analyze and visualize sensor data, in the form of wattage, across multiple circuits and recognize not only specific appliances but discrete events, such as a refrigerator or air conditioner turning on or shutting off. Homeowners can use such functions to check on their house while they are away, receiving alerts for events such as the sump pump turning on or simply checking to see if they remembered to turn the oven off. The Sense Home Energy Monitor has now been consistently rated as one of the top 5 home energy monitors for two years by several consumer review publications.