Survivable autonomy for the attritable era.

Mechanical layout

Software pipeline

Single-point altimeter

ToF proximity array

Full 3D non-repetitive scan

Why ToF array, not 2D scanner?

Slamtec RPLIDAR A1 ($99) is the obvious low-cost pick — but at 170g on a 700g drone, it's a 24% mass penalty in the worst possible location (top-mounted spinning).

Six VL53L5CX ToF modules at 3g each give 6-axis 8×8 range grids for 25g total — seven times lighter than RPLIDAR for equivalent proximity coverage on the axes that matter for evasion. Plus they run silently (no spinning motor), have no wear-out mode, and align directly with the 6-camera dome geometry.

Real part numbers, real stock

// AEGIS-FPV camera dome (forward EO)
// Bambu X1C / Prusa MK4 · PETG-CF
$fn = 96;
DOME_OD     = 28;        // mm outer
DOME_ID     = 24;        // mm inner
WINDOW_D    = 16;        // optical aperture
SHELL_T     = 1.6;       // wall
MOUNT_BOLT  = 3.2;       // M3 clearance
MOUNT_PITCH = 24;        // bolt circle

module dome() {
  difference() {
    sphere(d = DOME_OD);
    sphere(d = DOME_OD - 2*SHELL_T);
    cylinder(h = 30, d = WINDOW_D, center = true);
    translate([0,0,-15]) cube([40,40,30], center=true);
  }
}

module mount_flange() {
  difference() {
    cylinder(h = 3, d = MOUNT_PITCH + 8);
    cylinder(h = 9, d = WINDOW_D, center = true);
    for (i = [0:90:359])
      rotate([0,0,i+45])
        translate([MOUNT_PITCH/2,0,-1])
          cylinder(h = 6, d = MOUNT_BOLT);
  }
}

dome();
translate([0,0,-1]) mount_flange();

// AEGIS-FPV motor soft-mount (×4)
// TPU 95A · vibration isolation pad
$fn = 64;
ARM_W       = 22;        // arm width
ARM_T       = 4;         // carbon thickness
PAD_T       = 3;         // damping pad thickness
MOTOR_BC    = 16;        // M3 bolt circle (T-Motor 2207)
SHAFT_D     = 5.2;       // 5mm shaft + clearance

module pad() {
  hull() {
    translate([-12,0,0]) cylinder(h = PAD_T, d = 12);
    translate([ 12,0,0]) cylinder(h = PAD_T, d = 12);
  }
}

module mount() {
  difference() {
    pad();
    cylinder(h = 9, d = SHAFT_D, center = true);
    for (i = [0:90:359])
      rotate([0,0,i])
        translate([MOTOR_BC/2,0,-1])
          cylinder(h = 6, d = 3.2);
    translate([0,-ARM_W/2,-1])
      cube([ARM_W*2, ARM_W, ARM_T+0.4]);
  }
}

mount();

Jetson Orin NX carrier + sensor hub

1. Frame prep — debur carbon arms, M3 thread inserts pressed (Heat-Set TIM-3.0×4.0)
2. Motor install — torque to 0.25 Nm, run-in cycle (5 min @ 30% throttle, free spin)
3. ESC mate — 4-in-1 stack assembly, gyro damping pad, BLHeli_32 flash + ESC sync verify
4. FC + AEGIS-CARRIER — soldered standoff stack, conformal coat mask, HumiSeal 1B73 dip
5. Camera dome — 6-cam carrier press-fit, FFC routing, mechanical clearance check
6. Wiring + harness — terminate XT60, motor leads, signal loom, heat-shrink + flux clean
7. Power-on test — current draw <0.6A idle, voltage rails verified (5V/3.3V/1.8V), Jetson boot + ROS 2 launch verify
8. IMU 6-point cal — gyro/accel + mag in Helmholtz coil for BUF builds (figure-8 acceptable for non-BUF)
9. RF range bench — anechoic chamber 100mW, log link margin to 1.2km LOS
10. Camera cal — 6-cam intrinsic (K, D) + extrinsic (R, t) via checkerboard; stitched panorama check
11. Hover trim — 3-min station hover, log step responses, PID tune verify
12. ATP flight — 8-min mission profile, log review, tagging, QA sign-off

• ↗ PX4/PX4-Autopilot — flight stack
• ↗ ArduPilot/ardupilot — alt firmware
• ↗ bitcraze/crazyflie — open HW pattern
• ↗ NXP HoverGames — Altium files
• ↗ NVIDIA Jetson DG — DG-10931 carrier guide
• ↗ Holybro X500 V2 — PX4 ref vehicle
• ↗ Sierra Circuits MIL-PRF — DFARS PCB
• ↗ JLCPCB — fast-iter PCB
• ↗ PX4 uXRCE-DDS — FC↔companion bridge
• ↗ PX4/px4_msgs — ROS 2 contracts

AEGIS-S

• Compute: Jetson Orin Nano 8GB · 40 TOPS
• Sensors: 4-cam awareness dome
• Airframe: 5" / 580g AUW
• FC: Holybro Kakute H7
• Doctrine: Attritable · 1-way mission
• Volume: 1k MOQ · production-line

AEGIS-R

• Compute: Jetson Orin NX 16GB · 100 TOPS
• Sensors: 6-cam full-sphere dome
• Airframe: 5" / 700g AUW
• FC: Holybro Kakute H7
• Doctrine: Recoverable · multi-sortie
• Volume: 100 MOQ · qualification fleet

AEGIS-X

• Compute: Jetson AGX Orin 64GB · 275 TOPS
• Sensors: 6-cam + Livox Mid-360 LiDAR
• Airframe: 7" hex / 1.4kg AUW
• FC: Auterion Skynode S (BUF cleared)
• Doctrine: Persistent · manned-escort
• Volume: 10 MOQ · evaluation

Effective cost-per-mission falls 10× to 100× below current programs of record.

Defensive autonomy is a force multiplier on per-unit price. A platform that survives contact returns to the magazine; a platform that dies on the first salvo is consumed. AEGIS-FPV converts a single-mission attritable into a multi-sortie asset without altering its airframe class.

Detailed program economics, Replicator alignment math, and contract-vehicle guidance are in the Annex (see PDF brief). All cost figures cite primary DoD contract announcements published 2023–2025.

Photon to PWM

Typed topics