Welcome to our group's WTH Project: Bear-ly Awake 🐻!
In a world filled with smartwatches and clinical sleep trackers, we believe that the journey to better sleep should be comforting, not intimidating. Bear-ly Awake is a concept for an intelligent, interactive plush companion designed to help you achieve restorative sleep by addressing both the physiological and emotional aspects of rest.
Most modern solutions offer raw data without the human touch. Our project combines the warmth of a comforting object with the power of an AI-driven dashboard, creating a holistic experience that is as soothing as it is insightful.
-
Simulated Sleep Tracking:
- Users can start and end a sleep session, and the app will simulate and display live sensor data (Heart Rate, Temperature, and Movement etc.).
-
AI Sleep Assistant:
- A dedicated section provides a personalized routine based on the simulated sleep data, showcasing the core value proposition of an AI-driven companion.
-
Sleep History & Analytics:
- All simulated sleep sessions are logged and stored persistently. The dashboard provides an easy-to-use interface to review past sleep logs and analyze trends.
- Nicholas Wee (Group Leader)
- Ethan Ng
- Jacelyn Tan
- Ang Ke Jin (Jonathan)
- Rui Ahn
This repository contains the Streamlit-based MVP for the Bear-ly Awake concept. The goal of this MVP is to demonstrate the core user experience and validate the main features without the need for physical hardware. All sensor data is simulated, allowing us to focus on the user interface, data visualization, and the interactive flow of the application.
We hope this MVP helps you envision the future of Bear-ly Awake.