The KED – Kutta Edge Device is a ruggedized, wearable computing and networking hub designed to place processing resources close to personnel operating at the tactical edge. Its ARM-based Linux architecture draws 3–7 watts, limiting heat output and supporting extended operation while providing up to 200 eTOPS of AI performance. The device accommodates PyTorch, ONNX, and TensorFlow frameworks for edge-deployed artificial intelligence workloads.
KED supports automated target recognition, video compression, TAK services, automated Cursor on Target messaging, and data exchange across multiple platforms. Local processing reduces the volume of information that must traverse tactical networks, helping preserve situational awareness in connected, disconnected, and bandwidth-limited conditions. Its open, modular configuration enables integration with sensors, networks, and unmanned platforms while allowing capabilities to evolve over time.
| AI performance: | 200 eTOPS |
| Power consumption: | 3–7 W |
| Supported AI frameworks: | PyTorch, ONNX, TensorFlow |
| Architecture: | ARM-based |
| CPU: | 4-core A53 or 6-core A55 |
| Memory: | 4-core configuration: 8 GB LPDDR4; 6-core configuration: 16 GB LPDDR5 |
| Graphics: | 2D and 3D graphics acceleration |
| Video decode: | 4-core configuration: 1080p60 H.265, H.264, VP9, and VP8 |
| Video encode: | 4-core configuration: 1080p60 H.265 and H.264 |
| Integer performance: | Up to 16,560 DMIPS |
| Operating systems: | Yocto, Debian, Boot2QT, Android, FreeRTOS |
| External storage: | SD card support up to 2 TB |
| Net Warrior ports: | EUD, BATT, and two PAN ports |
| Display connection: | DisplayPort |
| USB: | USB 3.2 |
| Ethernet: | 10/100/1000 Mbps |
