Image Objects Using a DIY Configurable IR Tomography System

For his final year project at UCT (University of Cape Town), electrical engineering student Johnathan Whitaker designed an IR tomography imager using a ring of sensors and LEDs. According to Whitaker, he came up with CIRTS — Configurable Infra-Red Tomography System using simulation to come up with a functioning design tested his idea on a rotating platform and then built rings of sensors and LEDs to image objects. He can also take the garnered data and feed it to machine learning algorithms to identify objects or estimate their positions.

Whitaker’s first entry into IR tomography was a rotating scanner, which he designed using several 3D-printed components- including some gears and a base plate which houses a pair of stepper motors that turn said gears. A Teensy microcontroller drives the steppers, which positions a single phototransistor and LED at different points around an object to image it.

His next venture for the CIRTS project uses eight alternating phototransistors and LEDs arranged on a 3D-printed circular housing, which are also wired into a Teensy. Whitaker states admittedly that imaging objects were poor with this setup, but throwing some machine learning at the issue allowed him to estimate the location of an object down to around 3mm, which enabled him to differentiate objects with high accuracy. He used this setup as a game controller to steer ships and dodge enemies using a finger.

Whitaker is now in the process of building a bigger ring using 14 sensors in a non-symmetric arrangement that is expected to give him a higher resolution. Those that would like to develop their own CIRTS imaging system can use the code, files, and folders uploaded to Whitaker’s GitHub page.