Defense technology
pulled from the Æther
Independent research and development of autonomous systems, computer vision, and unmanned aerial platforms.
Autonomous systems Counter-UAS Visual navigation Celestial navigation NDAA compliant Edge AI

Built for the real world

Long-exposure photograph of a missile arcing across a dark night sky against stars
STRIX

Autonomous counter-drone
guidance system

Hardware-agnostic autonomous drone interception software that detects, classifies, and intercepts hostile unmanned aerial systems using onboard computer vision and proprietary guidance algorithms. The system identifies multiple target classes — including loitering munitions and commercial quadcopters — and autonomously closes on or maintains formation with the target without operator input. Designed to run on edge compute hardware connected to any ArduPilot-based flight controller, with full dual-stream telemetry logging per mission session.

Computer vision · Autonomous guidance · Edge AI · ArduPilot · MAVLink · Multi-class detection · Mission logging

Earth at night — city lights visible from orbit
PHANTOM

GPS-denied visual
navigation for drones

A visual odometry and map-matching navigation system that enables fully autonomous drone operations in GPS-denied and GPS-spoofed environments. The system ingests a live camera feed, extracts visual features in real time, and matches them against preloaded satellite imagery stored offline on the aircraft. Position estimates are fused directly into the flight controller's navigation filter, enabling stable autonomous flight across all major flight modes without any GPS signal present. Designed as a drop-in navigation layer for existing airframes.

Visual odometry · Map matching · GPS denied · Sensor fusion · Offline maps · ArduPilot · Edge compute

Hubble Ultra Deep Field — thousands of distant galaxies
SunStone

Autonomous celestial
navigation system

A two-node embedded navigation system that derives aircraft attitude and position from star field imagery — providing GPS-independent navigation in environments where both GPS and visual landmarks are unavailable. A dedicated capture node acquires star field images and forwards them over a low-latency link to a processing node running an AI-accelerated star pattern recognition model. The system solves celestial coordinates in near real time and outputs navigation solutions suitable for fusion with inertial sensors. Architected for future integration with inertial measurement hardware for full inertial-celestial hybrid navigation.

Celestial navigation · Star pattern recognition · AI accelerated · GPS independent · Embedded systems · Sensor fusion · Neural network

Illuminated fiber optic strands glowing blue against black
GOSSAMER

Tethered drone ground
unit architecture

Deep technical research into the internal electronics and signal architecture of fiber optic tethered drone ground units. The research covers the complete bidirectional signal path from the operator workstation to the airborne platform — including optical transceiver selection, wavelength multiplexing for single-fiber bidirectional communication, clock and data recovery, and power delivery over the tether. Provides a foundation for designing ground units capable of carrying simultaneous HD video, telemetry, and command streams over a single optical cable with minimal latency and no RF exposure.

Fiber optics · Bidirectional comms · Optical transceivers · Tethered UAS · Signal architecture · HD video · MAVLink

About Veilwork

Veilwork is an independent defense technology research and development organization focused on autonomous unmanned systems. We design and build software and hardware solutions that address real operational gaps — from counter-UAS guidance to GPS-denied navigation to alternative positioning systems. All work is conducted with NDAA compliance and real-world deployability as first-order requirements.

  • Autonomous systems

    Systems that close on targets across all major flight modes, without operator input.

  • Counter-UAS

    Onboard detection, classification, and interception of hostile unmanned aerial systems across multiple target classes.

  • Visual navigation

    Live camera feature extraction matched against offline satellite imagery for GPS-denied and GPS-spoofed environments.

  • Celestial navigation

    Aircraft attitude and position derived from star field imagery via AI-accelerated star pattern recognition.

  • NDAA compliant

    Compliance and real-world deployability treated as first-order engineering requirements, not afterthoughts.

  • Edge AI

    AI-accelerated perception and guidance models running entirely on airborne edge compute hardware.

Work with Veilwork

Defense technology research, autonomous systems development, and independent hardware consulting.