Loading cart...
The Sustainable AI Smart Lock is a high-tech security solution that turns recycled waste into cutting-edge hardware. Built by our engineering interns, this project features an entirely offline facial-recognition system powered by the Arduino Uno Q and Edge Impulse.
What if the plastic bottle you recycled last week became the lock securing your lab today? That's exactly what a team of engineering interns at Hardware Innovation Valley built — a fully offline, facial-recognition door lock whose entire physical housing was 3D printed from ocean-bound plastic bottles processed by PlastiBytes.
The project addresses a deceptively simple problem: traditional access methods fail. Keys get lost, passwords get compromised, and cloud-connected smart locks introduce new vulnerabilities. This lock sidesteps all of that. By running a lightweight neural network directly on an Arduino Uno Q microcontroller, the system performs facial recognition entirely on-device — no internet connection, no remote server, no single point of failure.
| Phase | What the Team Did |
| 1 — Train the AI | Captured face images across varied lighting and angles, trained a custom model in Edge Impulse, then deployed it directly to the microcontroller via Arduino App Lab. |
| 2 — Design the Lock | Designed a precision gear-and-rack deadbolt mechanism in CAD, then printed every component using PlastiBytes recycled PETG filament — proving sustainable materials can meet real mechanical demands. |
| 3 — Integrate & Demo | Wired a servo motor to translate AI decisions into physical movement. On demo day: 98% match confidence, a servo clicking to life, and a deadbolt sliding cleanly open. |
| Component | Tool / Material |
| Processing Brain | Arduino Uno Q |
| Vision | USB Camera Module |
| AI Training & Deployment | Edge Impulse + Arduino App Lab |
| 3D Printing Filament | PlastiBytes Recycled PETG |
| Actuation | Custom CAD Gear & Rack + Servo Motor |
This project makes two important points about where technology is headed:
Edge AI doesn't require data centers. Powerful, private, real-time inference is possible on a microcontroller that fits in your hand — no cloud, no subscription, no privacy trade-off.
The circular economy isn't a compromise. The recycled PETG filament handled the mechanical stress of a working gear system without issue. Sustainability and performance are not in conflict.
The project is fully open-source. CAD files and code are available at github.com/plastibytes/Face-ID