RoboCore

• 

  A layout of how the application is supposed to look like.

Inspiration

I wanted to create a bridge the hardcore 'robot' and an artificial brain.

What it does

RoboCore is a control hub that bridges the gap between high-level AI reasoning and physical robotic action. It can control an arduino to have an artificial intellighence of its own and it can also help a speaker to answer like a genAi.

How we built it

Frontend: Vite + React for high-speed performance.

Backend Integration: Firebase Functions to securely communicate with the Gemini API.

Communication: I designed a custom schema to translate AI text responses into "Action Commands" that a robot's hardware can actually execute.

Challenges we ran into

Since this was the first robot that i attempted to control using code, I had a hard time to tecah it how to move, walk, talk. But thankfully, as we were allowed to using the build feature of Gemini AI, it became a lot easier to train it over a period of time.

Accomplishments that we're proud of

Getting the Gemini API to reliably output structured data for physical movements was a "Eureka!" moment for me.

What we learned

I deepened my knowledge on Full- Stack Coding using Firebase, and also learned to build a robot that can interact with me using TinkerCard and some online Harvard Courses.

What's next for RoboCore

While experimenting with the codes, i burnt the arduino chip inside my robot. But I did not have enough timeline my chips and wires could be delivered before the submission deadline. So I need to repair it. I plan on taking my project a step further and develop it so it can act independently on wifi, without a cable attachment and train it to increase its response time and improve its actions.

Built With

• firebase
• gemini
• gemini-api
• google-cloud
• typescript