Neural Radiance Fields for the Real World: A Survey

2025-01-22

AI Summary

NeRF research suffers from fragmented theoretical advances, a lack of systematic cross-domain application analysis, and unclear practical deployment bottlenecks. To address these, this paper presents the first comprehensive survey of NeRF’s foundational principles—implicit neural representation, differentiable rendering, and multi-view geometry—and unifies over one hundred NeRF variants with their state-of-the-art applications in 3D reconstruction, visual understanding, and robot interaction. We establish a holistic landscape covering major datasets (e.g., DTU, Mip-NeRF 360), software toolchains (JAX/PyTorch implementations, training acceleration libraries), and evaluation metrics (PSNR, LPIPS, reconstruction completeness). Key challenges—including limited generalization, difficulty modeling dynamic scenes, and constrained real-time inference—are explicitly identified. Finally, we propose forward-looking research directions targeting embodied intelligence and edge deployment, delivering the first systematic roadmap bridging NeRF theory to real-world utility.

Technology Category

• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Applications
• Intelligent Robots: Other Foundations and Applications

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Web applications in cross-disciplinary domains and verticals such as mixed reality, smart cities, and digital health
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Neural Radiance Fields (NeRFs) have remodeled 3D scene representation since release. NeRFs can effectively reconstruct complex 3D scenes from 2D images, advancing different fields and applications such as scene understanding, 3D content generation, and robotics. Despite significant research progress, a thorough review of recent innovations, applications, and challenges is lacking. This survey compiles key theoretical advancements and alternative representations and investigates emerging challenges. It further explores applications on reconstruction, highlights NeRFs' impact on computer vision and robotics, and reviews essential datasets and toolkits. By identifying gaps in the literature, this survey discusses open challenges and offers directions for future research.

Problem

• Neural Radiance Fields
• 3D Scene Reconstruction
• Computer Vision and Robotics

Innovation

• Neural Radiance Fields
• 3D Scene Reconstruction
• Robotic Applications