This project aims to restore the balance of privacy and safety in environments where aerial surveillance leads to direct kinetic threats."
If you are struggling with FPV threats that DCNN/YOLO cannot detect beyond 500m, this is for you. We use SPAD to reach 5km. No AI bullsh*t, just pure physics.
"If the sky is watching, we are the ones who blink back."
Project Argus is a decentralized, open-source hardware initiative to build a solid-state optical radar. Our mission is to detect and neutralize intrusive UAV surveillance using the Cat-eye Retroreflection effect.
We don't just "see" drones; we lock onto their optical souls.
- 🟢 1.0km - 1.5km (The Shield):** Real-time, high-confidence detection of consumer-grade CMOS sensors. Instant lock-on.
- 🟡 1.5km - 3.0km (The Watcher):** Reliable tracking of professional/industrial UAVs using TCSPC algorithms.
- 🔴 3.0km - 5.0km (The Horizon):** Limit detection for early warning against long-range tactical surveillance.
This is a Solid-State Optical Array. No moving parts, zero CV latency.
- Emitter: High-power VCSEL Matrix (850nm) + Integrated Microlens Array (MLA) for ultra-collimated beam-steering.
- Detector: SPAD (Single-Photon Avalanche Diode) array with picosecond timing resolution.
- Logic: PRN (Pseudo-Random Noise) encoding + Time-Correlated Single Photon Counting (TCSPC) to pull signals out of solar noise.
- Anti-Surveillance: Coaxial optical path design allows for immediate laser dazzle/blind once a lens is confirmed.
We are looking for experts to move from Architecture to Action:
- Optics Wizards: Simulation of MLA focal planes and beam divergence at 5km.
- Hardware Hackers: High-speed PCB design for SPAD quenching circuits and VCSEL drivers.
- Signal Processors: FPGA/DSP experts to implement real-time TCSPC and noise-floor reduction.
- Embedded Fighters: ESP32/C++ masters for system integration and telemetry.
This project is licensed under CERN-OHL-S v2 (Strongly Reciprocal).
The Rule: If you use this shield, you share your improvements. No one owns the sky. We keep the defense open, forever.
- Check the
/Hardwarefolder for initial schematics. - Read the
/Whitepaperfor the link-budget calculations at 5km. - Join our developer group (Link in DM/Issues).
"Privacy is a human right. Physics is our weapon."
Skeptics may question the feasibility of a 5km range. As the architect, I am disclosing the Three Physical Pillars of Project Argus. We welcome global experts to verify these link budget assumptions:
Standard CMOS sensors fail at extreme ranges due to the inverse square law. We bypass this limit using SPAD (Single-Photon Avalanche Diode) technology.
- The Physics: A single returning photon is sufficient to trigger an avalanche current.
- Algorithmic Edge: By employing TCSPC (Time-Correlated Single Photon Counting), we don't look for an "image." We look for temporal correlation. Even if only a handful of photons return from 5km, as long as they match our encoded pulse sequence, the system will lock on.
- MLA (Microlens Array): Instead of crude bulk optics, we use integrated microlens arrays to compress the VCSEL divergence to <0.5 mrad.
- Coaxial Architecture: The emitter and receiver share the exact same optical axis. At 5km, even a 0.01° offset is fatal. Our coaxial design ensures "What we see is what we illuminate," eliminating parallax errors and maximizing retroreflective return.
- Nanosecond Gating: The receiver window is only active for a few nanoseconds, precisely timed to the pulse's Time-of-Flight (ToF).
- Narrowband Filtering: Utilizing 850nm/940nm interference filters (FWHM <10nm).
- Result: The system ignores incoherent solar noise, focusing exclusively on the "echo" of our own PRN-encoded signal.
Project Argus is currently in the Arch-Release Phase. We are recruiting the first wave of "Warriors" to bridge the gap between concept and silicon:
- Hardware Cell: Who can translate the conceptual SPAD bias circuit into a production-ready KiCad schematic?
- Algorithm Cell: Who can write the first PRN (Pseudo-Random Noise) correlation code for ESP32-S3 or FPGA?
- Optical Cell: Who can run a Zemax simulation to calculate the spot size (circle of confusion) of our MLA array at 5km?
Leave a comment in the Issues or submit your design drafts via Pull Request.
- License: This project is licensed under CERN-OHL-S v2. (See the
LICENSEfile for full text). - Reciprocity: If you modify the hardware or firmware, you MUST share the improvements under the same license.
- Disclaimer: Project Argus is for privacy defense and educational purposes only. Users are responsible for complying with local radio and laser safety regulations.
Click to expand the core physics behind 5km detection
The Physics: A Fusion of Digital Periscope & Active Cat-Eye Detection The core of Project Argus lies in a lethal combination of two optical principles, upgraded for the 21st century.
- The "Digital Periscope" Logic (Coaxial Alignment) Traditional surveillance is passive. Project Argus turns the table by using a Coaxial Optical Path.
The Principle: By aligning the VCSEL emitter and the SPAD detector on the exact same optical axis, we create a "Digital Periscope."
The Advantage: Whatever the system "sees," it is already "aiming" at. At a distance of 5km, even a 0.1-degree misalignment means missing the target by meters. Our coaxial architecture eliminates parallax error entirely, ensuring that the return signal from a tiny drone lens is captured with surgical precision.
- The "Active Cat-Eye" Logic (Retroreflection) Every camera lens—whether on a $500 DJI drone or a $10,000 telephoto setup—acts like a cat's eye.
The Phenomenon: When our encoded laser hits a remote CMOS sensor, the light is reflected back directly to the source along the incident path (Retroreflection).
The "Argus" Upgrade: We don't just look for a reflection; we use SPAD (Single-Photon Avalanche Diode) to detect individual photons returning from 5km away. By using TCSPC (Time-Correlated Single Photon Counting), we can identify a hidden lens even if it’s tucked behind a tinted car window or deep in a forest, as long as it has a line of sight.
🚀 Real-World Applications This isn't just for anti-drone warfare. Because the physics of retroreflection is universal, the PhotonShield can be deployed for:
Privacy Shields for Vehicles: Detecting dashboard cameras or tailing vehicles in real-time.
Industrial Counter-Espionage: Identifying long-range telephoto lenses aiming at sensitive factory floors from distant hills.
Autonomous Perimeter Security: A "set and forget" tile that alerts you the moment any optical sensor "looks" at your property.
To move from architecture to a functional Proof of Concept (PoC), we are prioritizing the following component selection. We invite hardware engineers and optical physicists to review and challenge these specs in the Issues section.
- Light Source: 850nm / 940nm High-Power VCSEL Array (Pulse Power >10W).
- Driver Topology: GaN-based FET drivers for sub-5ns rise times to ensure high spatial resolution.
- Collimation: Initial testing via aspheric condensers; target transition to Integrated Microlens Arrays (MLA) for <0.5 mrad divergence.
- Sensor: Exploring SPAD (Single-Photon Avalanche Diode) discrete components or CMOS-integrated SPAD arrays (e.g., STMicro FlightSense or Hamamatsu Silicon PMs).
- Filtering: Ultra-narrowband interference filters (FWHM <10nm) centered at the emitter wavelength to suppress solar background noise.
- Signal Path: High-speed Transimpedance Amplifiers (TIA) → Ultra-fast Comparators → FPGA LVDS inputs.
- Phase 1 (Validation): ESP32-S3 utilizing RMT/I2S peripherals for synchronized PRN pulse generation and basic gated windowing.
- Phase 2 (Scalability): Xilinx Zynq / Artix-7 FPGA for real-time Time-Correlated Single Photon Counting (TCSPC) and multi-channel correlation.
We are officially opening the following workstreams for contributors:
- Link Budget Analysis: We need a rigorous verification of the SNR at 5km considering the retroreflection cross-section of a standard 25mm CMOS lens.
- Coaxial Optical Housing: Designing a 3D-printable chassis that maintains sub-millimeter alignment between the emitter and the receiver.
- PRN Sequence Design: Optimization of Pseudo-Random Noise sequences to improve signal-to-noise ratios in high-ambient-light environments.
"The physics is sound. The components are available. Let's build the shield."
- High-Speed Scanning: FPVs move fast. We need to optimize the SPAD array's scanning frequency to ensure zero-miss in high-dynamic scenarios.
- Directional Alert: Integrate a simple compass/IMU module to provide a "Clock Position" (e.g., "FPV at 2 o'clock, 4.5km").