Problem Statement: Rapid at-home COVID Tests are not accessible to people with visual impairments due to the testing process requiring the participant to identify and distinguish different parts of the test using color, text, or position.
Two different engineering goals were identified for our designs. The first goal for our deliverable was conveying instructions for using the antigen testing kit in an accessible way for an individual with a visual impairment. The second goal was reading the results of a COVID-19 rapid antigen test and conveying them in an accessible way as well. In order to address these goals, the group made three preliminary prototypes. The first prototype focuses on communicating recorded instructions to the user through an Arduino-controlled circuit with buttons and a speaker. The second prototype attempted to read the results of a COVID-19 rapid antigen test using RGB color sensors controlled by an Arduino microcontroller. The third prototype was a machine learning model which attempted to read test results using image classification. This model was built using an Edge Impulse model and was deployed as a website for mobiles. The final prototype was constructed by improving and combining parts of the previous three designs. The final design uses a Raspberry Pi to convey recorded instructions based on the user's voice commands, which are recorded using the microphone attached to the Raspberry Pi. The design uses an improved version of the ML model from the third prototype to take pictures of an antigen test through the camera attached to the device and classify its result. The instruction and results are conveyed verbally back to the user through the speaker attached to the Raspberry Pi. By incorporating the previous three designs and improving upon them, the final design meets many more of the design criteria and is a more efficient and accessible deliverable.
Prototype #1: Instructions Device
Prototype #2: RGB Sensor
Prototype #3: Image Classification Model