Home » demoConstruct

demoConstruct

SLAM, Gaussian Splatting | Python, Babylon.js, WebXR

Overview

demoConstruct is an open-source project that blends SLAM, digital twins, and collaborative 3D scene editing. It goes beyond traditional 3D reconstruction to create an accessible and practical system for real-world applications in disaster response, medical rehabilitation, and collaborative design.

Key Features

• Progressive 3D Reconstruction: Real-time, accurate reconstruction of physical spaces with immediate visual feedback
• Collaborative Editing: Multi-user environment for simultaneous editing and interaction in shared 3D scenes
• Cross-Platform: Web-based interface, accessible via standard browsers on desktop, tablet, and VR devices

Technologies Used

• Front-end: Babylon.js, WebXR, Swift
• Back-end: Python, SLAM, Gaussian Splatting, MongoDB

Applications

• Medical Rehabilitation: Virtual environments for physical therapy and recovery
• Disaster Response: Rapid 3D mapping of affected areas for emergency planning
• Architectural Design: Collaborative design and visualization of architectural projects
• Industrial Training: Virtual training environments for complex industrial procedures

Presentations

• ACM CHI 2025: Showcased at CHI Interactivity 2025, Yokohama, Japan (April 26 – May 1).
• ACM SIGGRAPH 2024: Showcased at SIGGRAPH Labs 2024, Denver, USA (July 28 – August 1).

Project Links

• Project Website: demoConstruct
• GitHub Repository: immersification-demoConstruct
• Centre for Immersification: Research Center

Research Impact

demoConstruct advances spatial computing by combining SLAM, collaborative 3D editing, and accessible web-based tools. Its open-source framework enables researchers and developers to adapt the system across diverse domains, from healthcare to emergency response.

Currently, the platform is being piloted with Tan Tock Seng Hospital (TTSH), Singapore, to support virtual rehabilitation in clinical settings, demonstrating its potential for real-world applications.