A cyber-physical system to design 3D models using mixed reality technologies and deep learning for additive...

Area of Science:

• Computer Science
• Engineering
• Human-Computer Interaction

Background:

• Current 3D modeling for additive manufacturing often requires specialized skills and software.
• Integrating virtual and real-world interactions for design remains a challenge.
• Leveraging artificial intelligence (AI) and mixed reality (MR) offers potential for more intuitive design processes.

Purpose of the Study:

• To present advances in the I-nteract cyber-physical system for real-time 3D model generation.
• To enable automated 3D model creation from single 2D images using deep neural networks.
• To introduce a novel generative neural architecture, SliceGen, for diverse 3D model generation.

Main Methods:

• Development of a cyber-physical system (I-nteract) integrating MR technologies.
• Application of deep neural networks for inferring 3D models from 2D images.
• Integration of a novel generative neural architecture (SliceGen) to overcome limitations in 3D model genus.
• Implementation of user-adjustable dimension controls for workspace integration.

Main Results:

• Automated generation of 3D models from single 2D images using AI.
• Successful generation of diverse 3D models beyond single-genus limitations via SliceGen.
• Demonstration of furniture 3D model generation (chairs, tables) and placement in a physical space using MR.
• Validation of the system's effectiveness in a mixed reality environment.

Conclusions:

• The I-nteract system provides a novel, immersive interaction for 3D model design.
• The integration of AI and MR facilitates consumer involvement in personal fabrication.
• Advances in generative models like SliceGen enhance the capability for complex 3D model creation.