Novel sensing physics
We invent the measurement, not just the software around it — from optics and motion to signal geometry.
Perception · from first principles
We see what others can't.
A Priori Robotics builds sensing systems that start from physics, not catalog parts. Passive detection and 3D tracking of many drones at once — and a patent-pending depth camera that turns a single moving lens into a sub-millimeter close-range 3D scanner.
<3sto acquire & track
5+simultaneous targets in 3D
0–50cmdepth-sensing band
1moving camera, full depth
Scroll
The thesis
Two hard perception problems. Two clean solutions.
Most sensing hardware trades cost, range, or clutter for one another. We refused the trade-offs and re-derived each system from the ground up — a priori.
Live tracking
Airspace awareness
Passive multi-drone detection & 3D tracking
Detect, localize, and track multiple drones simultaneously — in seconds, in three dimensions. Fully passive: no emissions, no license, nothing for a target to detect or jam. A single sensor node builds a live 3D picture of everything in the volume.
- ✓Multi-target, from moment one. Simultaneous acquisition and 3D tracks — swarms don't hide in the count.
- ✓Passive by design. No RF emission to detect, jam, or license — you stay invisible while they don't.
- ✓True 3D. Range, azimuth, and elevation — not a 2D blip. Feeds directly into C2 and effector cueing.
Patent pending
Close-range 3D
Introducing the active parallax depth camera
Principled geometric depth from one camera. Active pure-translational motion generates motion parallax — and parallax is inversely proportional to depth. We directly track feature transit across the sensor and reconstruct a dense 3D point cloud in the near field — exactly where structured-light and stereo depth cameras struggle most.
- ✦Single sensor. The motion is the baseline. Motion combined with high-frequency sampling creates an effectively continuous baseline, avoiding the limitations of fixed-baseline multi-view geometry — no calibrated stereo pair, no projector, no feature search space, no feature-matching error, and no minimum distance.
- ✦Unrivaled near-field accuracy. Purpose-designed for 0–50 cm, where accuracy matters most for manipulation & inspection.
- ✦Physics, not guesswork. Depth from measured parallax — geometric, interpretable, and dense.
What we bring
Perception engineering, end to end.
Real-time 3D
Dense reconstruction and multi-target tracking that run live — engineered for the edge, not the datacenter.
Passive & covert
Emission-free architectures that observe without being observed — no RF signature to detect or defeat.
Deployable systems
From breadboard to fielded unit: we own the hardware, firmware, and perception stack top to bottom.
- <3s
- Acquisition to confirmed 3D track
- 5+
- Concurrent drone tracks per node
- 0–50cm
- Parallax depth working range
- 1×
- Camera required for full depth
Why "a priori"
Reasoning from the physics up.
A priori knowledge is what you can derive before you ever measure — from the structure of the problem itself. That's how we build. Before we choose a sensor, we ask what the physics actually permits, then engineer the shortest path to it.
The result is hardware that does more with less: fewer emitters, fewer cameras, fewer assumptions — and far more signal.
Derive the limit
What's the most information the physics of the scene can yield? Start there, not at the parts bin.
Design the measurement
Engineer motion, optics, and geometry to extract that information directly.
Field it
Collapse the stack into a rugged, real-time unit that ships.
Let's talk
See it detect. See it in 3D.
We run live demonstrations for defense, industrial, and research partners. Bring your hardest perception problem.