Inspiration
Imagine you're a student, mugging day in day out for examination. You look to your left, your phone has a notification. BAM, you're distracted. You think to yourself: 'This is an issue, I can't focus on my work!'.
Introducing BearWithIt, an interactive study companion that acts as both a teacher and a friend. BearWithIt is able to provide academic guidance, set a study routine like the Pomodoro technique, and can also listen to your problems. Having a physical representation of an AI companion, though much research has been done into the field such as Boston Dynamics or Stanford Robotics, has not been explored in an educational and commercial aspect enough. Many students use AI Chatbots like ChatGPT or Gemini to help them in their studies, but it is through a computer interface which can limit emotional connection.
What it does
Conversational AI. When talking to the bear, the Speech-to-Text API from Amazon Transcribe processes the microphone input in real time, and converts it to text. The built-in AI from Amazon Bedrock is able to process the text and give a response using Claude 3 Image Recognition. The bear is able to detect based on eye movement if the user is studying properly. If the user’s eyes close and is about to fall asleep, OpenCV detects the closed eyes and triggers a reminder to the user to focus Timer function. The bear is able to execute the Pomodoro technique, whereby users break work into intervals of 25 minutes, with 5 minute breaks in between.
How we built it
AWS Transcribe: Used for converting spoken language into text, allowing the bear to recognize verbal commands and questions accurately. This improves accessibility for users who prefer voice interaction. AWS Polly: Ensures the bear has a natural and engaging voice, creating a more lifelike and enjoyable interaction for users. AWS Bedrock: Leveraged for natural language processing, enabling the bear to understand and respond to user queries effectively. Its integration ensures conversational AI capabilities are reliable and scalable. OpenCV: Powers the face detection functionality, a critical component for monitoring focus and ensuring the bear can support productivity goals effectively.
Challenges we ran into
Integration: Initially planned to make the physical bear a standalone product, but the limited time and hardware availability made us rethink our mvp.
Accomplishments that we're proud of
We made our the functions for the mvp of the bear!
What we learned
Opencv, AWS and esp32 technologies
What's next for BearWithIt
Physical implementation of the model into the bear as a standalone product
Log in or sign up for Devpost to join the conversation.