Yerevan pilot · launching 2026

Waste, recognised.
Sorted in under two seconds.

HexaBin pairs dual OAK-D depth cameras with Google Gemini vision to classify urban waste into six material streams in real time — turning the act of sorting into a high-tech cultural ritual.

6
AI categories
2
OAK-D cameras
8
Planned locations
24/7
Always on
§ 01 · The project

Living
landmarks
for the smart city.

HexaBin is a revolutionary urban waste system that fuses IoT-driven Smart City technology with biophilic, nature-inspired design — turning a mundane chore into a civic ritual.

Deploying first in Yerevan, Armenia, every unit uses dual OAK-D depth cameras and a Gemini-powered vision model to recognise items into six material streams plus an "Empty" state — with brand-aware accuracy on Armenian products like Jermuk and BJNI.

Our bins act as visual landmarks in the urban fabric — using colour, material and small rewards to signal that the user is participating in a global movement of sustainability. We bridge the gap for citizens currently unmotivated to sort.

Classification accuracy
Coming soon
Armenian product benchmark in progress
Heartbeat interval
30s
edge → server
Sensor stack
Depth + IMU + NN
OAK-D native voting
Edge runtime
Raspberry Pi
Docker, restart-safe
§ 02 · The modular system

Six material streams,
one elevated language of colour.

Each module is visually intuitive yet architecturally sophisticated — premium tones reflecting urban civic space, not the trash room.

01 / 06

Paper

Warm Cellulose

Newspapers, cardboard, office paper, packaging. Sophisticated take on traditional fiber.

02 / 06

Aluminum

Brushed Metal

Beverage cans, foil, metal containers. Industrial circularity, detected by depth + vision.

03 / 06

Organic

Deep Biophilic

Food scraps, biodegradable packaging, garden waste — composted for circular sustainability.

04 / 06

Glass

Crystal Teal

Bottles, jars, glass containers — identified by their unique transparency signatures.

05 / 06

Plastic

Synthetic Tone

PET, HDPE, films. Classified by brand-aware AI trained on local Armenian products.

06 / 06

Other

Dynamic Gradient

The "brain" of the system. Adapts based on city-wide data, special collection, system signals.

§ 03 · Transparency by design

Every sort, measured.
Published from day one.

Each HexaBin logs every classification to the central fleet server. When the Yerevan pilot goes live, this section becomes a public real-time dashboard — updated every five seconds across all deployed units.

Items sorted

Every classification counted — per bin, per material stream.

Recyclable share

How much of the city's waste gets a second life, across the whole fleet.

CO₂ avoided (est.)

Emissions saved by diversion, modelled per material stream.

Fleet health

Live heartbeats from every unit — cameras, sensors, uptime.

Sort latency

From camera capture to sorting decision, measured end to end.

Live with the first deployment · 2026
§ 04 · Deployment map

Eight planned landmarks in Yerevan.

Planned · first wave Planned · second wave
YEREVAN · ARMENIA · 40.18°N 44.51°E
§ 05 · Media

See HexaBin in motion.

Loading 3D model…
3D MODEL
INTERACTIVE · GLB
3D product render
Industrial design study · drag to orbit
FOOTAGE
Coming soon
B-ROLL · 4K
Real-life deployment
Yerevan pilot, Q3 2026
DEMO
Coming soon
SCREEN CAPTURE
AI classification
Brand-aware vision in action