Post by GRNET - Greek Research & Technology Network

🚁 Scaling Digital Twins for UAVs with HPC Safe integration of UAVs into U-space requires more than simulations — it requires high-fidelity Digital Twins that capture the physics of wireless propagation, AI-based sensing, and autonomous flight dynamics at scale. To reach these scientific goals in the framework of EUSOME, our models must handle 38 million unknowns for ultra-massive MIMO propagation modeling and over 2 million constraints for UAV trajectory optimization. This is where High-Performance Computing on the ARIS infrastructure becomes essential. By leveraging ARIS Fat Nodes (512GB RAM) and the A100 GPU partition, our application is designed for high-performance scaling using: • iterative numerical solvers • memory-efficient structured representations • distributed workloads across nodes This enables capabilities that are simply not feasible on standard hardware: ⚡ TB/s-scale aggregate bandwidth across nodes 🧠 Stable training of high-resolution multimodal AI models (Vision + Radar up to 1024×1024) ✈️ Massive Monte Carlo validation of autonomous UAV flight safety 📊 Data scale of the workflow • Input: 3D urban meshes for ray tracing + high-resolution multimodal sensor datasets • Ray tracing output: benchmarking phase generating up to 5 TB (with production targets exceeding 50 TB) • AI training: 10 runs × 100,000 steps, requiring repeated reading of large-scale datasets • Flight optimization: 10⁵ Monte Carlo UAV scenarios, each satisfying >2 million constraints The result: high-fidelity UAV Digital Twins capable of validating sensing, communications, and flight safety in complex urban airspaces. This kind of HPC-enabled research is key to bridging the gap between theoretical models and real-world autonomous UAV operations. More👉https://bit.ly/41c044R EUSOME Project KIOS Research and Innovation Center of Excellence Athena Research Center Hellenic Drones S.A. Cyprus Chamber of Commerce and Industry Adrestia R&D Hellenic U-space Institute Centrum Badań Kosmicznych Polskiej Akademii Nauk SocialTech Lab Institute Mihajlo Pupin Hellenic Civil Aviation Authority #HPC #DigitalTwins #UAV #USpace #AI #WirelessCommunications #HighPerformanceComputing