I worked with Professor Giorgia Chinazzo of the CARE Lab to develop our own IoT sensor hub device to conduct living lab experiments.
I learned how to dive into unfamiliar technologies, collect and analyze large amounts of data and take a product idea from ideation and technical requirements to a functional prototype.
Project 1: Illuminance Sensor Comparison Experiment
There was a lack of research in the performance of low-cost illuminance sensors in relation to laboratory-grade, expensive sensors. Professor Chinazzo and I conducted this experiment to fill the knowledge gap.
For my first project in the CARE Lab, I conducted an experiment to test several affordable illuminance sensors against a laboratory-grade DeltaOhm PHOT-01 sensor. The objective was to find the most accurate illuminance sensor to use on a sensor hub device and to calibrate the output of the sensor to the laboratory-grade sensor.
I custom-built a stand and sensor holders to hold the laboratory-grade sensors and experimental sensors at the same horizontal plane to make light shine consistently across all of the sensors.
With the results of the illuminance sensor study, I developed a sensor hub device similar to UT Austin's Bevo Beacon device. This sensor hub will transmit data on Wi-Fi and will utilize the ESP32 microcontroller. I will also create a housing case to make occupants in the living lab spaces more at-ease around the sensor hubs.
We determined, through looking at the R^2 values of our linear regressions between the test sensors and reference sensors, that the TSL2591 sensor performed the best for values under 1000 lux and the TSL2561 performed the best for values over 1000 lux.
Project 2: Custom Sensor Hub Development
Commercial sensor hubs are expensive, don't consistently disclose the sensor model they are using, and don't give the researcher control over data cleaning and processing. Developing an in-house custom sensor hub gives the researcher full control over the hardware and software while keeping costs low.
I first researched the sensor models most commonly used in research papers, along with the sensor models used in other sensor hub projects. Using this data, Professor Chinazzo and I decided which sensor models to use in our own sensor hub.
I created a proof-of-concept WiFi datalogger built around an ESP32 to gather data from multiple sensors over several days. Coincidentally this prototype captured a humidity spike due to flooding in my apartment. Not a fun way to verify that it works!
I also led the ideation for the design of the sensor hub housing. Because this device will be placed close to the room's occupant (for example, on a desk), the aesthetics are important to the project. Lab members brainstormed widely, from dual-function designs to geometric art pieces.
Ultimately, we decided on a design resembling the Bosco Verticale, a residential building in Milan, Italy that incorporates greenery in the facade. Professor Chinazzo was a fan of the natural aesthetic and the added bonus of the green shrubs hiding the sensors from view, decreasing the sense of surveillance.
I finished my time at the CARE Lab by creating an MVP functional prototype that transmits data to Google Sheets via Wi-Fi with a structural 3D-printing housing. The next steps from here are to develop a PCB to remove the wires and to create manufacturing plans for the housing based on the design of the Bosco Verticale.
