Jointly established by Beijing Institute of Architectural Design and an experienced digital twin team, we focus on the digital twin simulation data infrastructure needed for the implementation of spatial intelligence and embodied intelligence. Leveraging the top design team’s experience in complex spaces, engineering standards and industry scenarios, as well as our team’s capabilities in digital twins, physical simulation and synthetic data generation, we are committed to building engineering-accurate and physically credible digital twin environments available for training, alongside an integrated closed loop covering simulation data collection, model evaluation and physical prototype verification for embodied intelligence.
Supporting over 100 3D model formats, a 3D twin scene, cloud rendering + WebGL dual engines, full coverage across large screens, desktops, and XR/VR devices, while meeting the demands of high-concurrency access for ultra-large-scale twin scenes and film-level high-fidelity rendering. Leveraging a physics engine to achieve physically accurate behavior and operation simulation of twin scene objects. Fully web browser-based, requiring no software or plugin installation, and providing complete JavaScript and TypeScript interfaces for seamless and rapid integration with your business systems.
Using the Multiverse Engine, 3D models in over 100 formats can be converted into Sim Ready simulation assets that meet simulation requirements. These simulation assets can be directly imported into mainstream simulators such as Isaac Sim/Lab, Gazebo, and MuJoCo. Our research paper on the simulation asset production process has been accepted by the ICLR 2026 conference. Additionally, the Sim Ready asset library we contributed to has been included in Nvidia's officially recommended Isaac Sim asset library. Meanwhile, our Synthesis asset platform has open-sourced over 15,000 high-quality simulation assets to date, including a large number of items with articulated joints.
Digital Twin Asset Platform
Construct 1:1 mapped digital twin assets of physical assets and integrate them with 3D model objects, drawings, IoT devices, video, and other data. Leverage the Multiverse Engine to rapidly build digital twin business scenarios, and utilize external data interfaces to drive asset objects, achieving accurate simulation of production, construction, and operational processes, as well as equipment maintenance. Built-in knowledge graph + AI large language model (optional for private deployment or integration with online services) automatically understands and generates asset topology diagrams, equipment maintenance guides, and knowledge graphs for various professional documents, enabling intelligent Q&A.
