Streaming Physical AI Digital Twins and Immersive Experiences to XR Headsets with NVIDIA CloudXR 6 on AWS

This article presents a new AWS Sample for running NVIDIA CloudXR 6 on Amazon EC2 GPU instances, enabling GPU-rendered immersive experiences to stream to XR headsets over the network with minimal latency.

• NVIDIA CloudXR 6 streams GPU-rendered OpenXR content to XR devices, supporting Apple Vision Pro and WebRTC clients like Meta Quest and Pico headsets
• Physical AI applications include digital twin reviews, simulation validation, and teleoperation for robot data collection in realistic 3D environments
• Reference architecture uses Amazon CloudFront for entry point, Amazon ECS signaling proxy with Amazon Cognito authentication, and EC2 GPU instances in AWS Local Zones for low-latency streaming
• Testing achieved approximately 35 ms pose-to-render latency at 90 FPS with g7e instances, delivering seamless immersive experiences
• GitHub sample includes architecture documentation, deployment guides, and a minimal demo for quick setup

The solution enables teams to access photorealistic, physics-accurate simulations and digital twins from lightweight XR headsets without requiring local GPU workstations.