Breakthrough in Safety-Critical AI Computing for Tactical Platforms to be showcased at AFCEA 2026

Zug, Switzerland, April 2026

As defense and aerospace applications increasingly require more intelligence at the edge, ELMA, Wind River, Klepsydra Technologies, and AEI are collaborating on a high-performance, secure, and SWaP-optimized computing architecture. The solution is designed to support demanding tactical workloads and will be demonstrated at the AFCEA Conference in Bonn, Germany, in May 2026.

The Solution: Consolidating the Modern Tactical Stack

The deployment of AI-driven applications often requires heavy ruggedized computers with a high power consumption. This new ecosystem focuses on high performance, integration velocity, security hardening and long-term support on a compact edge computer.

To demonstrate the power and versatility of this architecture, the partners will showcase an Unmanned Ground Vehicle (UGV) use case, where the system simultaneously manages:

• Autonomous Maneuver: Execution of mission-critical perception, obstacle detection and path planning algorithms.
• System Health Analytics: Real-time Predictive Maintenance (PdM) monitoring mechanical components, such as bearing systems, for early indications of failure.

Integrated Technical Architecture

The strength of this solution lies in the seamless integration of four industry-leading technologies:

• Ruggedized Processing (ELMA): A high-performance computer built for harsh environments, centered on the NVIDIA Orin AGX to provide massive GPU density.
• Optimized Linux Platform (Wind River): Wind River Linux is carefully tuned to deliver low-latency performance in a small footprint GPU-accelerated edge computer. Optimized for NVIDIA Orin platforms, this OS enables rapid integration of CUDA-based AI frameworks, high-throughput inference pipelines, and long-term maintainability.
• AI Acceleration (Klepsydra Technologies): A high-performance inference engine that maximizes GPU throughput, allowing multiple AI models to run in parallel without latency spikes.
• Logic & Simulation (AEI): Safety-trained AI models and high-fidelity simulation data that provide the intelligence for complex autonomous behaviors.

Significance for Military Modernization

This architecture represents a scalable path for deploying “Trusted AI” across a variety of platforms, from ground vehicles to airborne systems. By reducing the physical footprint of the computing stack—often referred to as Size, Weight, and Power (SWaP) optimization—the solution enables more agile and capable next-generation maneuver platforms.

“We are proving that high-performance AI inference and SWaP optimization are no longer mutually exclusive,” the partners stated. “Our unified architecture provides a mature, cross-sector solution for the modernization priorities of international defense forces.”

Live Demonstration at AFCEA Bonn

Attendees are invited to visit the demonstration at AFCEA 2026 to see the hardware in action. The exhibit will feature a live simulation of a UGV navigating complex terrain while performing real-time mechanical self-diagnostics, all powered by a single NVIDIA-based ruggedized unit.

About the Partners:

ELMA

Innovative developer of ruggedized hardware and modular electronics.

Wind River

A global leader in delivering software for mission-critical intelligent systems.

Klepsydra Technologies

Specialists in high-performance AI middleware for edge computing.

Interested in learning more?

If you are attending AFCEA Bonn 2026, we invite you to visit us at the NVIDIA booth or schedule a meeting with the team via the link provided.