In preparation for the Fall 2020 semester, MIT needed to take urgent action to address COVID-19 on campus. To do this, MIT chose to create self-serve "kiosks" to help monitor vital signs of the large MIT population on a continuous basis.
The minimum viable product is a system to measure external body temperature to within FDA guidelines (< 0.5 degrees Celsius). We also want the system to:
Be scalable, so more sensors can be added after implementing on campus
Accommodate people over a range of heights (approx. 4ft to 7ft)
Be deployable in up to 50 locations on campus
Cost less than $10,000 per system
My work on the kiosk system was to research options for the minimum viable product. The resulting product was a Raspberry Pi 4 equipped with two sets of cameras. The camera set had a FLIR Lepton camera (for thermal measurement) and a Raspberry Pi camera module (for facial feature detection and image alignment). The camera sets were spaced at a distance so that anyone within the designed height range will be in the field of view of one of the two cameras. The sensors and electronics cost less than $1,500 per system, and the full system costs less than $5,000.
Using a Raspberry Pi as a central processor allows the system to be inexpensive while providing a flexible platform for other sensors to be added. Data acquisition is achieved using Python libraries, which are readily available for many sensors.
The following images demonstrate the different proposed workflows with the camera. Each option has trade-offs related to signal processing complexity and burden on the user.