AuraShield

Wiring diagram

Inspiration

High-profile individuals face constant privacy invasion. Paparazzi, stalkers, and unauthorized photographers threaten their safety and peace. Traditional security reacts after the damage is done — photos already taken, privacy already violated.

We built AuraShield: an AI-powered privacy defense system that detects cameras and neutralizes them instantly — without harming anyone.

What it does

1️⃣ Detection Viam camera + computer vision identifies:

Camera lenses (phones, DSLRs, drones)
Suspicious positioning/aiming behavior
Known threat patterns

2️⃣ Classification

Detected	Response
Camera aimed at property	IR flood activation
Drone with camera	Alert + tracking
Authorized person	Ignored

3️⃣ Neutralization

IR LED array floods the area with infrared light
Cameras capture only white-out — photos ruined
Human eyes see nothing — completely safe
AI voice warns: "Photography is prohibited on this property"

How we built it

Hardware:

High-power IR LED array (850nm wavelength)
Viam-connected camera for lens detection
Servo-mounted targeting system
Raspberry Pi controller

Software:

Viam — Hardware abstraction + cloud monitoring
OpenCV — Camera lens detection (reflective glint detection)
Google Gemini — Dynamic warning generation
Eleven Labs — Authoritative voice alerts
Python — Control logic

The Math Behind AuraShield

1. Sensor Saturation (The Neutralization)

Digital camera sensors (CMOS/CCD) respond to both visible and near-infrared light. The recorded signal $S$ at each pixel is:

$$S = \int R(\lambda) [I_{visible}(\lambda) + I_{IR}(\lambda)] d\lambda$$

Where:

$R(\lambda)$ = Sensor spectral sensitivity
$I_{visible}$ = Visible light intensity
$I_{IR}$ = Infrared light intensity

When AuraShield activates, we ensure $I_{IR} \gg I_{visible}$. This forces the signal to reach its maximum bit-depth:

$$S \geq S_{max}$$

2. Human Safety (The Stealth)

Human vision is limited to $400\text{ nm} \le \lambda \le 700\text{ nm}$. Because we use 850nm IR, human retinal sensitivity is effectively zero:

$$R_{human}(850\text{ nm}) \approx 0$$

The perceived brightness $S_{human}$ remains unchanged, ensuring the system is invisible:

$$S_{human} \approx \int R_{human}(\lambda) I_{visible}(\lambda) d\lambda$$

3. Lens Detection (The "Glint" Effect)

Camera lenses are multi-element curved glass systems that produce a characteristic specular reflection modeled by:

$$I_{glint} \propto I_{emit} \cdot \rho \cdot \cos(\theta)$$

Where:

$\rho$ = Lens reflectivity coefficient
$\theta$ = Angle relative to the camera axis

4. Tracking Math (The Lock-On)

To maintain a lock on moving paparazzi, we use a discrete-time kinematic model to predict the camera's position:

$$x_{t+1} = x_t + v_t \Delta t + \frac{1}{2} a_t (\Delta t)^2$$

Target Market

Celebrity homes (LA, SF, NYC)
High-net-worth residences
Corporate executives
Luxury hotels & venues
Private events

Challenges we ran into

📸 Lens Detection — Camera lenses create distinctive "glint" reflections. We trained detection on this optical signature.
🎯 Tracking — Paparazzi move fast. Predictive tracking keeps IR aimed correctly.
⚡ Power — High-power IR LEDs need proper thermal management.