---

Sub-domains

Sub-domain	Focus	Papers
3D Gaussian Splatting	3DGS fundamentals, optimization, rendering, compression	150+
Neural Radiance Fields	NeRF architectures, acceleration, generalization	100+
NeRF-SLAM	NeRF-based simultaneous localization and mapping	60+
GS-SLAM	Gaussian Splatting SLAM systems	50+
LLM-3D	Large Language Model-driven 3D understanding	60+
Robotics	Manipulation, navigation, locomotion, localization	80+

---

Key Research Areas

3D Gaussian Splatting

Real-time radiance field rendering via learnable 3D Gaussians. Covers the original 3DGS paper and extensions for dynamic scenes, text-to-3D generation, compression, large-scale scenes, and mesh extraction.

Key works: 3D Gaussian Splatting (SIGGRAPH 2023), GaussianEditor, DreamGaussian, SuGaR, Scaffold-GS, GaussianPro

Neural Radiance Fields (NeRF)

Implicit neural scene representations for novel view synthesis. From the original NeRF through Instant-NGP, Mip-NeRF 360, and extensions for text-to-3D, generative models, and large-scale reconstruction.

Key works: NeRF (ECCV 2020), Instant-NGP (SIGGRAPH 2022), Mip-NeRF 360, Nerfstudio, NeuS, NeuralAngelo

NeRF-SLAM & GS-SLAM

Integration of neural implicit/explicit representations with simultaneous localization and mapping for real-time 3D reconstruction. Both visual and LiDAR-based approaches.

Key works: iMAP, NICE-SLAM, Co-SLAM, SplaTAM, MonoGS, Gaussian-SLAM

LLM-3D Understanding

Large language model-driven 3D scene understanding, grounding, and reasoning. Connecting text-based reasoning with spatial 3D representations.

Key works: 3D-LLM, LEO, LL3DA, Chat-3D v2, Scene-LLM, Grounded 3D-LLM

Robotics

Applying 3D vision to robotic manipulation, navigation, locomotion, and localization. NeRF/GS-based approaches for robot learning and sim-to-real transfer.

Key works: NeRF-RL, GNFactor, Ditto, SplaTAM-based navigation, 3D-aware manipulation

---