Conventional ISR video sensors force a tradeoff between coverage and detail. Zoom in for identification and lose awareness. Zoom out for area and miss threats. The ECHO Camera System eliminates that compromise, a 100MP sensing front end, onboard AI processing, and real-time dissemination in a single payload stack from 700g.
Three core elements work as one. ECHO captures 10,000 x 10,000 pixel frames at full operational resolution. FORGE processes them onboard and in real time. IGNITE runs your AI models at the edge, converting raw sensor data into geolocated, CoT-formatted detections — streamed to operators in the field without high-bandwidth data links.
ECHO:10K
ONE FRAME.
EVERYTHING IN IT.
ECHO is the sensing front end of the ECHO Camera System. A 100MP CMOS sensor with electronic global shutter and native Canon RF lens mount in a compact 60 x 60 x 65mm package. Field-swappable lenses to match a broad range of mission profiles - from wide-area surveillance at 24mm to high-altitude tactical reconnaissance at 300mm.
A single PCIe link to FORGE carries imagery, power, control, and hardware sync. No separate harness, no external connectors on the sensor. Mount ECHO on the gimbal; route compute wherever thermal and weight budgets allow. RGB and Monochrome sensor variants cover 400–950nm for visible and Near-IR missions.
FORGE
GPU INFERENCE.
AT MISSION FRAME RATES.
FORGE is the onboard processing and integration module. It runs the IGNITE:AI edge framework, stores captures, and exposes every bus your UAV needs, Ethernet, MAVLink, CAN, USB — so the ECHO Camera System integrates as a single coherent payload, not a sensor tethered to a generic compute board.
Built on NVIDIA Jetson Orin NX with 16GB, FORGE delivers the GPU compute required for 100MP inference at mission frame rates. 2TB onboard SSD. 190 x 55 x 30mm at 160g — designed to mount inside sealed enclosures, pods, or airframe bays without external penetration.
IGNITE
FROM RAW PIXELS
TO GEOLOCATED INTELLIGENCE.
IGNITE is the edge AI framework that turns raw sensor data into actionable intelligence on FORGE. It handles everything upstream and downstream of your model: sensor preprocessing, GPU tiling, smart cropping, georeferencing, object tracking, and C2-ready output formatting. You supply AI model in a Docker container. That is the boundary.
Deploy your own trusted AI models, PyTorch, TensorFlow and ONNX compiled to TensorRT for inference on FORGE. IGNITE extracts full-resolution crops from 100MP frames for each detection, assigns persistent tracking IDs across frames, and formats findings as Cursor-on-Target events for direct dissemination to ATAK and existing C2 systems.
