Digital Twin Replication for Cross-Reality Simulation Engines

Abstract

Digital twins—precise, virtual replicas of physical systems—have rapidly evolved into indispensable instruments for capturing, simulating, and optimizing real-world assets and processes. As immersive technologies spanning augmented reality (AR), virtual reality (VR), and mixed reality (MR) converge into the broader concept of cross-reality (XR), the demand grows for digital twin frameworks that can seamlessly integrate across heterogeneous simulation engines. This manuscript introduces Digital Twin Replication for Cross-Reality Simulation Engines (DTR-XRSE), a comprehensive architecture designed to achieve bidirectional synchronization between physical entities and their virtual counterparts within AR/VR/MR environments. Our approach comprises a layered architecture—spanning sensor data acquisition, edge processing with real-time filtering and compression, cloud-native orchestration for dynamic simulation-instance management, and XR client integration via low-latency WebRTC streams. We detail the data model standardization, MQTT-and-Kafka messaging pipelines, Kubernetes and K3s orchestration strategies, and Unity 3D XR integrations that underpin DTR-XRSE. A proof-of-concept implementation involving a six-DOF industrial robotic manipulator demonstrates the framework’s capability: sub-5 ms round-trip latency, sub-0.5 mm positional error, and 25 % improvement in XR training task completion times versus baseline systems. We also analyze throughput scaling across multiple edge nodes, autoscaling behaviors under Kubernetes, and user-experience metrics drawn from the Igroup Presence Questionnaire (IPQ).