Project Heimdall — Full Specifications

Everything inside the helmet that sees everything

Project Heimdall — Three market configurations revealed

Apex Rider HUD — Prototype Testing

12 integrated subsystems. One helmet.

100

Helmet Shell

Modified full-face shell, EPS liner minimum 15mm, polycarbonate or fiber composite outer, all modifications preserve structural integrity.

200

Single-Board Computer

NVIDIA Jetson Orin Nano 8GB, 1024-core Ampere GPU, 32 Tensor cores, 40 TOPS AI inference, 256GB NVMe SSD via M.2.

300

Multi-Camera Array

4 cameras: primary forward 12MP 90-degree chin bar, two lateral 8MP 160-degree temple, rear 12MP 170-degree fisheye crown. Sub-25ms end-to-end pipeline latency via hardware-accelerated ISP.

400

MR Display Assembly

Two Kopin Lightning micro-OLED displays, minimum 1280×720 per eye, minimum 3000 nits, positioned at interpupillary distance.

500

Motorized Lens Mechanism

Two MG90S servo actuators, pivot arms, Adafruit I2C PWM driver, 0.8s deploy travel, 0.6s retract travel, 15G crash safety interlock.

600

Aerodynamic Cooling

Passive intake vents at chin bar high-pressure zone, copper heat spreader, internal duct, exhaust vents at rear crown low-pressure wake zone. Active 5V blower below 30mph. Maintains SBC below 55°C at highway speed.

700

WiFi CSI Through-Wall Detection

Two ESP32-S3 nodes, anterior and posterior positions, 150mm+ baseline separation. Detects humans up to 5 meters through solid walls. 17 skeletal keypoints. Breathing and heart rate. Passive — uses ambient WiFi only.

800

Power Management

10,000mAh 3S 11.1V LiPo, BMS board, 5V and 12V regulated rails, three power modes: Standby 5W, Active 10–15W, Emergency beacon only. 2.5 to 3 hours runtime at Active mode.

900

HUD Software Layer

OpenGL hardware-accelerated compositor. Speed and RPM via Bluetooth OBD-II. Rear view picture-in-picture. Blind spot alerts. Navigation overlay. WiFi human detection overlays. AI threat alerts. Lens cleaning status.

1000

AI Threat Recognition

YOLOv8 TensorRT weapon detection at 15fps minimum, confidence threshold 0.70. Optional ArcFace facial recognition submodule, disabled by default, local watchlist only, no external transmission.

1100

Night Vision

Hardware: 850nm/940nm IR illuminator array plus IR-sensitive sensor without IR-cut filter. Software: neural low-light enhancement model on SBC GPU, zero additional hardware required. Mode selectable via HUD or physical switch.

1200

Lens Cleaning

Hydrophobic and oleophobic coating on all lenses. Micro air-knife with miniature pump and nozzle per camera. Piezoelectric vibration element per lens housing. Image-quality monitoring algorithm scores each camera independently and triggers escalating cleaning sequence automatically. No liquid reservoir.

Protected by 32 patent claims

Apex 1080 has filed USPTO Provisional Patent Application No. 64/108,615 covering 32 claims that protect the integrated architecture of Project Heimdall — from the multi-camera passthrough pipeline to the automated lens cleaning system. This provisional application establishes priority date for the full non-provisional utility patent filing.

NVIDIA Jetson Orin Nano — 40 TOPS AI Sub-25ms end-to-end latency BOM under $1,500 USPTO App. No. 64/108,615

Interested in Heimdall for your organization?

Contact Us