© 2019 by Manisha Mohan

Excuse My Posture

2015

Collaborator: Andres Calvo

Excuse my posture is a chair that can determine if users are sitting with good or bad posture with 14 force sensitive resistors on its surface. After describing the characteristics of good posture, a software tool prompts the user to alternate between sitting in good and bad posture while collecting data. The software uses this data to train a machine learning classifier (namely, a support vector machine) that is personalized for the user. Afterwards,the system uses the classifier to evaluate the user's posture in real time. When the user has bad posture, the system provides subtle ambient feedback by gradually blurring the computer screen until the user returns to good posture. Similarly, if a user is listening to music, the ambient feedback instead distorts the music by mixing it with Gaussian noise. The relative amplitude of the Gaussian noise gradually increases as a user remains in bad posture, and the noise is removed as soon as a user returns to good posture.