Columbia University EECS E4764 IoT Project Report #0

Posture Helper

It is a system which could help people develop healthy

sitting posture and eye care habits without physical therapy

Motivation

Nowadays, most people spend more than 8 hours sitting at the desk, and bad posture could result in disease like chronic fatigue syndrome, cervical spondylosis, heart and breathing related problems.

Different Postures

System

Architecture

Embedded System: we use IR distance sensor to obtain data. Five behind user, working together to define a sitting posture. One in front of the user to prevent near sight.

Sensors and OLED Screen

Android app: Our app can show user's sitting habit with plots in detail.

Cloud:use the machine learning model trained before to judge what kind of posture it is from the data posted by Huzzah.

Accuracies for different machine learning models

Technical Components

IR distance sensor

Prototype

Traditional wearable posture corrector brace may have bad influence on dressing aesthetics, and don’t offer personalized correcting function.

Results

The OLED screen displays the user's realtime posture.
The first line shows the user's sitting posture. The second lines tells if the user is stting too close to the screen.