Skip to content
#edge computing Open access

Daimon Interceptor

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)

Abstract

Based on the provided documentation, here is a comprehensive summary of the automated interceptor system's architecture, divided into its core subsystems and physical requirements. Core Functional Disciplines An automated short-range or counter-unmanned aerial system (C-UAS) interceptor is split into three foundational engineering pillars: Detection & Sensor Fusion: Employs long-range Active Electronically Scanned Array (AESA) radar for initial target tracking, transitioning to terminal homing seekers (Global Shutter Optical Cameras with edge AI, millimeter-wave radar, or uncooled infrared). An Extended Kalman Filter (EKF) continuously fuses sensor and IMU data to predict the target's future position. Guidance Laws: Utilizes Proportional Navigation (PN) to steer toward a predicted collision point. The system uses a Navigation Constant (N, typically 3 to 5), the Line-of-Sight Rate (Ω), and the Closing Velocity ($V_c$) to calculate commanded acceleration. Kinematics & Actuation: Implements precise physical course corrections via aerodynamic fin control (grid fins or canards driven by brushless DC servos) and Thrust Vector Control (TVC) / Reaction Control Systems (RCS) side-thrusters for rapid adjustments or high-altitude operations. Component Breakdown & Compute Stack The electronic core of an open-architecture bench prototype is organized into four main layers: Subsystem Key Components Primary Function 1. Compute Stack • NVIDIA Jetson Orin Nano (8GB)• Arducam OV9281 Global Shutter Camera• 6mm/8mm M12 optics Runs quantized INT8 object detection pipelines at low latency; global shutter prevents motion blur during fast transitions. 2. Flight Control • Teensy 4.0 (or STM32H7)• Bosch BNO055 9-axis IMU• TF-Luna LiDAR Module Functions as the autopilot running navigation loops at 400Hz+; calculates PN guidance updates and proximity metrics. 3. Actuation • SUZLAZYR 2804 Brushless Gimbal Motors• AS5600 / AS5048A Magnetic Encoders• Foxeer Reaper F4 Mini 4-in-1 ESC Translates guidance math into physical fin positioning using ultra-smooth Field Oriented Control (FOC). 4. Power & Infrastructure • 4S to 6S LiPo Battery (100C+)• Step-Down Buck DC-DC Converters Distributes high-current power for actuation while regulating clean voltage rails for the compute ecosystem. Airframe Platforms & Global Infrastructure Beyond the electronic brains, the vehicle requires integrated structural, communication, and power interfacing frameworks to become fully operational: Airframe Variables: The structural and propulsion framework changes depending on the mission profile: Rocket-Powered Projectiles use solid-fuel motors (APCP) and grid fins; High-Speed Multi-Rotors utilize high-KV brushless motors and carbon-fiber propellers; and Fixed-Wing configurations use EPP foam wings with pusher propellers for extended loitering ranges. Power Isolation: The Jetson Orin Nano demands a strict, high-amperage 12V or 19V dedicated buck converter to satisfy its 15W–25W draw. This must remain entirely isolated from the 5V logic/sensor regulator to prevent brownouts caused by motor current spikes. Signal Interfacing: Bi-directional logic level shifters (e.g., TXB0104) are required to safely translate UART or I2C data streams between the Jetson's 1.8V/3.3V logic and the Teensy's 3.3V/5V environments. Ground Control Link: Uses a long-range wireless telemetry link (such as ExpressLRS 2.4GHz/915MHz) paired with ground control software (QGroundControl or Mission Planner) for calibration, live tracking metrics, and manual safety overrides. Critical Engineering Realities for Assembly To avoid hardware failure during bench testing, three specific design rules must be applied: Thermal Control: Running continuous neural network tracking forces the Jetson to maximum power, requiring an official active heat sink and fan mapped via Linux PWM configurations to prevent thermal throttling. Vibrational Isolation: High-frequency FOC motor noise causes sensor jitter in the IMU. This must be mitigated using low-durometer silicone damping rings or gel tape, alongside a software-based low-pass filter under 50Hz. Magnetic Shielding: High-current wires and motor magnets disrupt magnetometers and encoders. The telemetry IMU must be physically offset by at least 5 to 10 cm using non-magnetic standoffs. Now that you have the complete architectural summary, what would be most helpful next? I can provide: The Teensy 4.0 C++ boilerplate code to parse target vectors from the Jetson. A physical schematic layout mapping the XT60 power splitting and buck regulators. A logic level wiring checklist to ensure your 1.8V, 3.3V, and 5V pins are safely isolated.

View source

Similar papers

#computer vision Review Sep 2017

Agile Software Development Methods: Review and Analysis

This publication proposes a definition and a classification of agile software development approaches and analyses ten software development methods that can be characterized as being "agile" against the defined criterion.

P. Abrahamsson, O. Salo, Jussi Ronkainen et al. · 727 citations · ⚡54
#computer vision Jun 2008

The impact of agile practices on communication in software development

The study shows that agile practices improve both informal and formal communication, but indicates that, in larger development situations involving multiple external stakeholders, a mismatch of adequate communication mechanisms can sometimes even hinder the communication.

M. Pikkarainen, Jukka Haikara, O. Salo et al. · 401 citations · ⚡48
#machine learning Review Open access Oct 2014

Software development in startup companies: A systematic mapping study

The results indicate that software engineering work practices are chosen opportunistically, adapted and configured to provide value under the constrains imposed by the startup context.

Nicolò Paternoster, Carmine Giardino, M. Unterkalmsteiner et al. · 394 citations · ⚡54

Related blog posts

Microsoft Research Blog Oct 6, 2026

What AI gets wrong and what failure teaches us

Jennifer Neville did not want to go into computer science—but that’s exactly where she landed. Neville discusses the starts and stops that led to her professional sweet spot and her work identifying “surprising failures” making it hard for AI to handle complexity.  The post What AI gets wrong and what failure teaches us appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.