Inspiration

I was deeply moved by meeting a visually impaired couple in Indonesia. The husband survived by begging on the streets, while the wife walked a treacherous morning route alone... Navigating broken infrastructure and open drops just to give massages for a few bucks. Globally, 338 million people face similar struggles. When I discovered that existing assistive smart glasses cost between $3,000 and $6,000, equivalent to a full year’s salary for the average worker in my home country, Indonesia. I realized high-tech independence was locked behind a massive wealth barrier. So I built ZoraLens to turn safety and independence from a high-priced luxury into an affordable reality.

What it does

ZoraLens is an open-source, modular assistive wearable device that provides semantic environmental description for the visually impaired. Using a low-cost camera module mounted on 3D-printed frames, the glasses capture the user's exact perspective, then routes it to a smartphone via a Type-C cable, and sends the image to the cloud-based Gemini AI to return a rich description of the surroundings straight to the user's phone and delivered to the user using text-to-speech (TTS).

How we built it

The core architecture is built entirely on a highly efficient, low-cost framework:

1. Hardware: The eyes of the system is a $22 Seeed Studio XIAO ESP32S3 Sense module. It is enclosed in a custom 3D-printed frame designed for effortless local assembly, bringing the total raw hardware production cost to just $26.
2. Software & AI: Offloading heavy compute requirements to the Gemini AI Cloud API and leveraging the user's existing smartphone, the ZoraLens application handles image routing to the Gemini AI Cloud API, generating descriptive natural language text and outputs it in the form of an audio using text-to-speech(TTS)
3. Accessibility Interface: The system maps the camera trigger to the phone's physical volume rocker button, ensuring users do not have to fiddle with touchscreens to scan their environment

Challenges we ran into

The journey was a steep learning curve. When testing the initial prototype with local users, I didn't even know how to 3D print yet and had to craft the physical frames out of cardboard. Dealing with a four-to-six-second network processing latency during cloud AI queries was a technical hurdle, but user feedback showed they prioritized clear environmental understanding and durable hardware over raw speed. Designing a custom 3D model that could house the electronics tightly while remaining simple enough for local, non-expert technicians to build was a significant iteration challenge. Furthermore, back in the early stages of development, I ran into a substantial problem when trying to implement my "modular" philosophy. I attempted to utilize a webcam to act as the "eyes" of the system, but failed to integrate it due to its power-hungry nature, making it incompatible with the smartphone-gateway system. However, I decided to pivot to a less power-demanding camera, which is the XIAO ESP32S3 Sense. This was a significant pivot as it preserved both the cost-efficient philosophy and the modular philosophy in which I've dreamed of..

Accomplishments that we're proud of

1. Engineered a 99% Cost Reduction: I successfully designed, sourced, and built a functional assistive wearable device with a total hardware bill of materials of just $26, proving that life-changing technology doesn't have to carry a multi-thousand-dollar price tag.
2. Overcame Early Hardware Bottlenecks: After hitting a major roadblock with a power-hungry webcam, I successfully pivoted the entire system architecture to the low-power XIAO ESP32S3 module, preserving both the device's mobility and my modular design philosophy.
3. Moved from Cardboard to Custom 3D CAD: I didn’t let a lack of formal training stop me. When my first physical prototype had to be taped together out of cardboard, I taught myself 3D modeling from scratch to design a durable, custom enclosure that can be built anywhere.
4. Conducted Real-World Field Testing: I am incredibly proud that I took my rough, early-stage prototypes directly into the community to test them with the visually impaired individuals who inspired the project. Hearing the first real-time semantic descriptions successfully play through the smartphone speaker proved the core concept works in practice.

What we learned

I learned that high-end polish isn't what matters most to communities in need. Instead, practical utility, accessibility, and empathy do. I also learned how to design hardware with a focus on product longevity. Instead of creating another disposable electronic device, I leaned heavily into a triplet structure of open source code, modular architecture, and decentralized repair so that the tech can outlast fragile commercial options.

What's next for ZoraLens

As a high school student founder, my immediate focus is on ground-level execution, community validation, and keeping the technology accessible:

1. Localized Pilot Testing: My next step is to partner directly with local chapters of NGOs, such as the Indonesian Blind Association (Pertuni), to run a small-scale pilot test with 3-5 users to measure daily device durability and gather feedback.
2. Open-Source Documentation: I am finalizing my GitHub repository to open-source the code and providing step-by-step 3D-printing and assembly guides so that caregivers can build and repair the device locally for $22 without a factory.
3. Student Workshops: I plan to introduce this "AI-augmented engineering" framework to my peers by hosting student workshops, helping others build affordable, functional hardware that addresses local community challenges.

tech4compassion

Built With

• android
• android-studio
• c++
• esp32
• gemini
• google-generative-ai-sdk
• jetpack-compose
• kotlin