This survey examines how Large Language Models (LLMs) can be enhanced with 3D spatial understanding capabilities, potentially outperforming traditional computer vision methods.

• Explores integration of LLMs with 3D understanding for robotics, autonomous vehicles, virtual reality, and medical imaging
• Proposes a comprehensive taxonomy of LLM-based 3D spatial reasoning approaches
• Identifies key challenges and opportunities at the intersection of language models and spatial cognition
• Highlights engineering applications in robotics and autonomous navigation systems

For engineering teams, this research opens pathways to develop more sophisticated autonomous systems that can understand and interact with complex 3D environments through natural language interfaces.

How to Enable LLM with 3D Capacity? A Survey of Spatial Reasoning in LLM