The Vault

Assembly
Landing Page
Values
CAD
Hardware Stack
Software Stack

Inspiration

Homes are getting smarter. Locks are getting smarter.
But safes and vaults are still fooled by photos, copied fingerprints, and stolen passwords.

In the age of AI, deepfakes, and identity theft, we realized something scary:

Most home security verifies data.
Almost none verifies a real, living human.

So we asked:
What if a vault could know it’s actually you?

That idea became The Vault.

How this fits the Home Improvement Theme

When people think about home improvement, they think about smart lights, smart thermostats, and smart locks.

But one of the most overlooked parts of a home is how we protect the things that matter most:

Passports
Jewelry
Important documents
Hard drives
Family valuables

Ironically, while homes are getting smarter, safes and vaults are still protected by outdated methods that can be bypassed with:

Photos
Copied fingerprints
Stolen passwords
Deepfakes

The Vault reimagines what a modern home safe should be in the age of AI.

Instead of verifying credentials, it verifies a real, living human being physically present.

This turns a traditional vault into an identity-aware home security system — not just for safes, but for doors, cabinets, server rooms, and any access point inside a home where identity truly matters.

That is a direct improvement to home security infrastructure — not just adding a lock, but redefining how access is granted throughout the home.

What it does

The Vault is a 4-layer biometric home security system that authenticates biological presence, not just credentials.

It requires:

Face recognition
Anti-spoof liveness detection (LiDAR / depth)
Fingerprint verification
Hashed keypad password

Even if someone has your password, a copied fingerprint, or a deepfake of your face — the vault will not open.

Because it doesn’t check what you have.
It checks who is physically standing there.

How we built it

We built a full hardware-software authentication pipeline:

React Frontend for the user interface
Flask backend as the authentication server
ROS2 to coordinate sensor and hardware modules
Face recognition + anti-spoofing for liveness detection
Fingerprint sensor + keypad connected via Arduino
PySerial for Arduino ↔ backend communication
Database for biometric templates and hashed credentials

System Architecture Overview

The Vault is built as a distributed hardware–software system using ROS2 for coordination.

Nodes

face_recognition_node (C++)
- Face detection and recognition using dlib
- Depth-based spoof detection using camera data
- Publishes authentication and spoof status to ROS topics
web_interface_node (Python / Flask)
- User management interface
- REST API for frontend
- Bridges authentication flow between sensors and UI

Frontend

React web interface for registering users and managing authentication

Hardware Integration

Arduino connected fingerprint sensor and keypad
PySerial communication between Arduino and backend
ROS2 orchestrates all modules into a single authentication pipeline

Challenges we ran into

Synchronizing Arduino hardware with backend logic in real time
Managing reliable serial communication between sensors and the server

Accomplishments that we're proud of

Built a working multi-biometric vault prototype end-to-end
Successfully integrated hardware, backend, and frontend into one cohesive system
Implemented anti-spoofing liveness detection, not just face matching
Designed a true biometric fusion authentication flow
Created a system that feels like a real product, not just a demo