Endurance
Minutes, not shifts.
Most small multirotors fly for under an hour per battery. Covering a site around the clock still means someone on hand to swap packs.
Illustrative simulation
Lamassu Systems · Drone technology
Detect. Charge. Fly.
We’re building the ground layer for drones that work around the clock: sensors that find and track small aircraft, pads that recharge drones without a human, and unmanned systems designed to tie it all together.
01 The problem
Drones got smart. The ground around them didn’t.
Modern drones can fly themselves. What they still can’t do is stay up, stay aware, or stay unattended for long. The bottleneck isn’t the aircraft — it’s the infrastructure on the ground.
Awareness
Small, low, and quiet.
Small drones are hard to spot: they fly low, blend into clutter, and may not broadcast anything at all.
Labor
A crew for every flight.
Launching, landing, charging, and watching the sky still take people on site — the cost that keeps drone programs from scaling.
Lamassu is building the missing ground layer: sensors that see small drones, pads that recharge them without a human, and systems that turn both into persistent operations.
02 The platform
One ground layer. Three capabilities.
Each capability stands on its own. Together they let drones operate continuously — sensing the airspace, returning to charge, and relaunching without a person on site.
Illustration · Pipeline demonstrated on prototype hardware
Detection & tracking
See what’s flying.
Passive, camera-first detection on edge hardware. Our prototype finds small drones in live video, keeps them tracked over time, and physically follows them with a pan-tilt mount.
- Real-time edge inference
- Multi-frame visual tracking
- Closed-loop pan-tilt follow
- Passive sensing — emits nothing
Concept illustration · In R&D
Autonomous charging
Keep them flying.
Landing pads that recharge drones without a human: precision landing, a hands-free connection, and battery health checks between missions.
- Vision-guided precision landing
- Hands-free charging interface
- Battery health telemetry
- Pad-to-fleet scheduling
Concept illustration · Exploring
UAS platforms
Built to come home.
Unmanned aircraft systems designed around the ground layer: onboard edge compute for perception, docking-ready hardware, and autonomy that knows when to return and recharge.
- Onboard perception
- Docking-ready airframes
- Mission autonomy
- Fleet operations software
03 Persistent autonomy
Launch. Patrol. Return. Recharge. Repeat.
Persistence is a loop, not a flight. Lamassu is designed so every stage of that loop can run without a person on site — while the same sensors keep watching the sky.
- LaunchA charged drone lifts off the pad on schedule or on alert.
- Patrol & senseIt flies its mission while ground sensors track everything else in the air.
- Return & landWhen the mission or battery says so, it returns and lands with visual guidance.
- Charge & checkThe pad recharges it hands-free and checks battery health before the next sortie.
04 Proof
Not a slide deck. A working prototype.
Our detection and tracking stack already runs on real edge hardware. Here’s what it does — and exactly how we measured it.
94.8 percent
Precision
95.1 percent
Recall
96.9 percent
mAP@50
Single-class YOLOv8n detector, measured on the validation split of a ~22,000-image public drone dataset — images held out from training. These are engineering validation results, not a guarantee of field performance in every condition.
How we measured it- ComputeNVIDIA Jetson Orin Nano edge module
- DetectLive camera inference with latest-frame capture
- TrackBoT-SORT multi-object tracking with persistent IDs
- FollowTwo-axis pan-tilt mount with calibrated limits
- NetworkMulti-node corroboration logic built; field tests next
05 Why now
The rules, the risks, and the compute just lined up.
Drone operations are moving from case-by-case exceptions toward routine infrastructure — and the public record shows it.
Aug 2025 · FAA proposed rule
Routine BVLOS is being written into the rules.
Directed by the FAA Reauthorization Act of 2024, the FAA’s proposed beyond-visual-line-of-sight rule is intended to provide
“a predictable and clear pathway for safe, routine, and scalable UAS operations.”
Jun 2025 · Executive order
Airspace awareness became national policy.
The order directs federal agencies to detect, track, and identify drones, and calls for guidance to help critical-infrastructure operators do the same.
Now · Edge compute
Real-time vision fits on the edge.
Embedded GPU modules now run real-time computer vision on compact, low-power hardware — the same class of compute that already runs our detection prototype on an NVIDIA Jetson Orin Nano.
06 The name
Named for a guardian.
In ancient Assyria, winged Lamassu stood watch at city and palace gates. We’re building the modern version: a quiet guardian at the edge of every site — watching the sky and keeping the fleet ready.
07 Build with us
Build the ground layer with us.
We’re looking for the people who need this first: sites with airspace to watch, fleets that need to stay in the air, and builders of the hardware that makes it possible.
- 01Pilot sites
Operators with airspace to watch or drones to keep flying.
- 02Hardware partners
Airframe, battery, charging, and sensor makers.
- 03Investors
Backers of infrastructure for persistent autonomy.
- 04Engineers
Robotics, embedded, computer vision, and power electronics.