# Query-based Semantic Gaussian Field for Scene Representation in Reinforcement Learning

arXiv: https://arxiv.org/abs/2406.02370 · HTML: https://mlsanity.com/paper/2406.02370/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in RL for robot locomotion and manipulation, one of 15 directions in Deep reinforcement learning methods.
- Builds on 42 of its 52 references that are in the map.
- First posted on arXiv 4 Jun 2024.

## Placed in

- Deep reinforcement learning methods / [RL for robot locomotion and manipulation](https://mlsanity.com/deeprl/rl-for-robot-locomotion-and-manipulation/)

## Builds on (most cited first, 30 of 42)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains](https://mlsanity.com/paper/2006.10739/) (2020)
- [pixelNeRF: Neural Radiance Fields from One or Few Images](https://mlsanity.com/paper/2012.02190/) (2020)
- [Offline Reinforcement Learning with Implicit Q-Learning](https://mlsanity.com/paper/2110.06169/) (2021)
- [Off-Policy Deep Reinforcement Learning without Exploration](https://mlsanity.com/paper/1812.02900/) (2018)
- [4D Gaussian Splatting for Real-Time Dynamic Scene Rendering](https://mlsanity.com/paper/2310.08528/) (2023)
- [CURL: Contrastive Unsupervised Representations for Reinforcement Learning](https://mlsanity.com/paper/2004.04136/) (2020)
- [MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereo](https://mlsanity.com/paper/2103.15595/) (2021)
- [What Matters in Learning from Offline Human Demonstrations for Robot Manipulation](https://mlsanity.com/paper/2108.03298/) (2021)
- [Block-NeRF: Scalable Large Scene Neural View Synthesis](https://mlsanity.com/paper/2202.05263/) (2022)
- [Learning Complex Dexterous Manipulation with Deep Reinforcement Learning and Demonstrations](https://mlsanity.com/paper/1709.10087/) (2017)
- [IBRNet: Learning Multi-View Image-Based Rendering](https://mlsanity.com/paper/2102.13090/) (2021)
- [HexPlane: A Fast Representation for Dynamic Scenes](https://mlsanity.com/paper/2301.09632/) (2023)
- [LangSplat: 3D Language Gaussian Splatting](https://mlsanity.com/paper/2312.16084/) (2023)
- [DeepVoxels: Learning Persistent 3D Feature Embeddings](https://mlsanity.com/paper/1812.01024/) (2018)
- [Local Implicit Grid Representations for 3D Scenes](https://mlsanity.com/paper/2003.08981/) (2020)
- [DISN: Deep Implicit Surface Network for High-quality Single-view 3D Reconstruction](https://mlsanity.com/paper/1905.10711/) (2019)
- [ManiSkill2: A Unified Benchmark for Generalizable Manipulation Skills](https://mlsanity.com/paper/2302.04659/) (2023)
- [Decoupling Representation Learning from Reinforcement Learning](https://mlsanity.com/paper/2009.08319/) (2020)
- [BungeeNeRF: Progressive Neural Radiance Field for Extreme Multi-scale Scene Rendering](https://mlsanity.com/paper/2112.05504/) (2021)
- [A Survey on 3D Gaussian Splatting](https://mlsanity.com/paper/2401.03890/) (2024)
- [NeRF-Editing: Geometry Editing of Neural Radiance Fields](https://mlsanity.com/paper/2205.04978/) (2022)
- [Behavior From the Void: Unsupervised Active Pre-Training](https://mlsanity.com/paper/2103.04551/) (2021)
- [Deep Spatial Autoencoders for Visuomotor Learning](https://mlsanity.com/paper/1509.06113/) (2015)
- [NeuMesh: Learning Disentangled Neural Mesh-based Implicit Field for Geometry and Texture Editing](https://mlsanity.com/paper/2207.11911/) (2022)
- [3D Neural Scene Representations for Visuomotor Control](https://mlsanity.com/paper/2107.04004/) (2021)
- [NeRF-DS: Neural Radiance Fields for Dynamic Specular Objects](https://mlsanity.com/paper/2303.14435/) (2023)
- [ManiGaussian: Dynamic Gaussian Splatting for Multi-task Robotic Manipulation](https://mlsanity.com/paper/2403.08321/) (2024)

## Built on by (most cited first, 6 of 6)

- [GAGS: Granularity-Aware Feature Distillation for Language Gaussian Splatting](https://mlsanity.com/paper/2412.13654/) (2024)
- [Feed-Forward 3D Scene Modeling: A Problem-Driven Perspective](https://mlsanity.com/paper/2604.14025/) (2026)
- [Humanoid Occupancy: Enabling A Generalized Multimodal Occupancy Perception System on Humanoid Robots](https://mlsanity.com/paper/2507.20217/) (2025)
- [RLGS: Reinforcement Learning-Based Adaptive Hyperparameter Tuning for Gaussian Splatting](https://mlsanity.com/paper/2508.04078/) (2025)
- [Learning Structural Latent Points for Efficient Visual Representations in Robotic Manipulation](https://mlsanity.com/paper/2605.21258/) (2026)
- [Deep Learning Perspective of Scene Understanding in Autonomous Robots](https://mlsanity.com/paper/2512.14020/) (2025)

## Abstract (arXiv)

Latent scene representation plays a significant role in training reinforcement learning (RL) agents. To obtain good latent vectors describing the scenes, recent works incorporate the 3D-aware latent-conditioned NeRF pipeline into scene representation learning. However, these NeRF-related methods struggle to perceive 3D structural information due to the inefficient dense sampling in volumetric rendering. Moreover, they lack fine-grained semantic information included in their scene representation vectors because they evenly consider free and occupied spaces. Both of them can destroy the performance of downstream RL tasks. To address the above challenges, we propose a novel framework that adopts the efficient 3D Gaussian Splatting (3DGS) to learn 3D scene representation for the first time. In brief, we present the Query-based Generalizable 3DGS to bridge the 3DGS technique and scene representations with more geometrical awareness than those in NeRFs. Moreover, we present the Hierarchical Semantics Encoding to ground the fine-grained semantic features to 3D Gaussians and further distilled to the scene representation vectors. We conduct extensive experiments on two RL platforms including Maniskill2 and Robomimic across 10 different tasks. The results show that our method outperforms the other 5 baselines by a large margin. We achieve the best success rates on 8 tasks and the second-best on the other two tasks.