Inspiration

In India alone, millions of blind and low‑vision people still rely almost entirely on a white cane or a human guide, which limits when and where they can move independently. Watching this daily reality in our own city made the team ask a simple question: what if assistive navigation technology could be as affordable and wearable as normal spectacles, instead of costing ₹1-2 lakhs like most imported devices?

What it does

“The Third Eye” is a pair of AI‑powered smart glasses that combines a Raspberry Pi Zero 2W, camera, GPS and microphone to guide blind users through the world hands‑free. The system detects obstacles in front of the user, tracks GPS location, understands simple voice commands, and responds with clear audio instructions plus haptic vibration alerts on the frame so the user always knows when to turn or stop.

How we built it

We flashed Raspberry Pi OS Lite onto a Pi Zero 2W, wired the Pi Camera, NEO‑6M GPS module, microphone and vibration motor, and brought them together in Python. OpenCV and a lightweight YOLO‑based model handle real‑time obstacle detection, while GPS and custom navigation logic estimate distance and direction to the destination. A voice layer built with Speech Recognition and pyttsx3 let's the user say commands like “navigate to the main gate” and receive spoken guidance. Finally, we designed a TPU 3D‑printed frame to hide the electronics inside a comfortable, glasses‑like form factor instead of a lab prototype.

Challenges we ran into

Working on a Pi Zero 2W forced careful optimization of the vision model so it could run in real time without overheating or draining the battery. Getting reliable GPS locks in dense areas and synchronizing sensor data with voice feedback required multiple rounds of testing. As a mostly software‑focused team, learning practical electronics—stable wiring, power management, and mechanical mounting inside a thin frame—was another big learning curve.

Accomplishments that we're proud of

We moved from idea to a working end‑to‑end prototype that can detect obstacles, speak turn‑by‑turn instructions and vibrate to warn the user, all running fully on‑device without cloud connectivity. The prototype is built using low‑cost, locally available components and a 3D‑printed frame, which keeps the estimated unit cost around a few thousand rupees—far below typical commercial assistive glasses.

What we learned

This project taught us how to design full IoT systems: from flashing OS images and configuring headless SSH access, through sensor integration and ML model deployment, to mechanical design and user‑centric testing. We also learned how important it is to involve real users early—simple decisions like where to place the button or how strong the vibration should be make a huge difference in usability.

What's next for The Third Eye

Next steps include collecting feedback from blind users in real environments, refining the obstacle‑detection model for Indian streets, and shrinking the electronics into a slimmer, production‑ready PCB. In parallel, we plan to add multilingual support for Indian languages and work with local NGOs to pilot “The Third Eye” with early adopters who can help us shape it into a product that truly fits their daily lives.