Bluebird’s proprietary real-time edge AI platform.
Most security estates are assembled from four vendors who never intended to work together — a camera company, a drone company, a video-management company and whoever integrates them. Bluebird builds the whole estate: airframes, cameras and gateways on one proprietary platform, manufactured in San Jose, California and Bangalore, India. One vendor, one software layer, one set of data.
BB Drone — Border Security, Surveillance & First Responder
The security and first-response airframe. Launches from a dock, holds station over an incident, and streams a classified scene rather than raw video — detection runs onboard, so the decision does not wait on a network.
- Max flight time90 min
- Max payload3.25 kg
- Wind resistanceLevel 9
- Max horizontal speed12 m/s
- Hover accuracy±0.1 m RTK
- Video transmission5 km
- Flight ceiling120 m
- Setup timeunder 1 min
- Thermal — ZT6 — 640 × 512, −20 to 550°C, ≤50 mK, 1080p visible
- Optical — 3-axis EO gimbal — 4K UHD, 20–30× optical, starlight sensor
- Multispectral — 5 bands, 2–5 MP each, NDVI and NDRE output
BB Drone — Last-Mile Delivery
The cargo airframe. Carries to the point of need and hands off at a secure drop zone, where the recipient unlocks with a phone. Folds down for transport between sites.
- Max flight time70 min
- Max payload1.4 kg
- Take-off weight5.2 kg
- Dimensions495 × 495 × 435 mm
- Folded300 × 390 × 430 mm
- Wind resistanceLevel 6
- Video transmission5 km
- Hand-offPhone unlock
Robin — Outdoor Edge AI
The workhorse fixed camera. Detection and classification run in the housing, so what leaves the camera is an understood event rather than a video stream.
- Resolution5 MP, 2592 × 1944
- Edge chipsetAmbarella CV75S
- Lens2.88–13.7 mm
- PowerPoE, max 3.4 W
- ModelBB-5VWD-GS
Sparrow — Outdoor HD IP Camera
Bullet-form outdoor camera for long perimeter runs and low light, rated for impact and weather exposure.
- Resolution5 MP, 2880 × 1620
- IR distanceup to 60 m
- RatingIP66, IK10
- PowerPoE, max 11 W
- ModelBB-5VB1-SO+
Skycatcher
Multi-functional camera for human and vehicle recognition. Built for car parks and urban transport roads, reading people and vehicles at speed with the AI stack running on the camera itself.
- ConfigurationDual lens
- Compute20 TOPS NPU
- Processor6-core ARM Cortex-A78AE
- Lens6–22 mm vari-focal, AF 4×
- Pan / tilt60° / 60°
- Shooting distance6 m, up to 20 m wide
- Vehicle speedunder 80 km/h
- RatingIP65, max 22 W
X-Series — X1 and X2
The site gateway. Aggregates every camera and airframe on the estate, runs the heavier cognition workloads locally, and keeps the whole picture on site.
- AcceleratorNVIDIA T1000 / Axelera Metis
- Compute8 GB GPU / 214 TOPS
- Memory16 / 32 GB DDR5
- Streams16 ch FHD
- RatingIP65, max 22 W
The Hornet platform and strong IP stack.
Every device reports into the same platform. Perception runs in the camera, cognition runs at the gateway, and the console shows understood events rather than footage to be searched.
Real-time security alerts
Events raised the moment they are classified, not after review.
Cross-camera insights
One object followed from camera to camera, and from camera to drone.
Customer insights
Demographics, occupancy and journey — without facial recognition.
Surveillance dashboard
Mapping the complete perimeter and within — assets online, drones in flight, and every alert on one live site map.
Logistics dashboard
Complete mapping and tracking from A to B — deliveries in flight, fleet payload state, and the drop zone confirmed on aerial imagery.
Footfall analytics
Heatmap and path flows across the floor — where people go, where they dwell, and how the layout converts.
Customer insights
Demographics and occupancy, attributed without facial recognition.
Real-time alerts
Cross-camera insights — every fixed camera plotted against the site, each pin holding the detections behind it.
Vehicle security analytics
Count, colour, make, model and plate — read at the edge, matched across every camera. Allow and deny lists alert at the gate.
An interconnected platform.
Air and ground assets hold one continuous track between them. A camera detects at the fence line, the gateway keeps the live model of the site, and a drone takes the same object over the moment it leaves fixed coverage — no re-detection, no gap at the hand-over, and one identity for the object from first sighting to last.
BB Drone · AI Camera
device-to-device mesh
video · events
X1 / X2
edge inference before uplink
inference
Aggregation, retention, reaction
aggregation · retention · reaction
Device and comms layer
- Device-to-device mesh, air and fixed assets
- Device-to-gateway telemetry and command
- Gateway-to-cloud uplink
Management layer
- Flight tasking and mission sequencing
- Asset health, availability and hand-off
- Operator command and override
Data layer
- Edge inference on device, before uplink
- Cloud aggregation across sites and assets
- Storage, retention and redundancy
Topology and flows above are drawn only from architecture Bluebird already asserts — edge processing, the X1/X2 gateway, and cloud aggregation. Protocols, bandwidth, latency and failover behaviour are deliberately not stated.