Research map / 3D vision / NeRF and radiance fields
SNeRL: Semantic-aware Neural Radiance Fields for Reinforcement Learning
Dongseok Shim, Seungjae Lee, H. Jin Kim
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use SNeRL encoder instead of DrQ-v2, CURL, CURL-multiview, CNN-AE, NeRF-RL | SNeRL with SAC · Window-open, Hammer, Drawer-open, Soccer (Meta-world, 4 seeds) | episode return | Improved |
| Pre-train with plain CNN autoencoder reconstruction (CNN-AE) | CNN-AE · Meta-world visual control tasks | episode return | Hurt |
| RGB-only NeRF autoencoder (NeRF-RL) versus multi-view contrastive CURL-multiview | NeRF-RL · Meta-world visual control tasks | episode return | No clear effect |
| Remove semantic supervision or distilled feature supervision (ablation) | SNeRL · Window-open and Soccer | episode return | Hurt |
| Use only semantic labels in the simple Window environment (ablation) | SNeRL · Window-open | episode return | No clear effect |
| Add multi-view self-predictive auxiliary loss (ablation) | SNeRL · Meta-world tasks | episode return | Mixed |
| Use SNeRL encoder in Dreamer instead of NeRF-RL or CNN-AE weights | Dreamer · Meta-world tasks, model-based | episode return | Improved |
| Add semantic and feature supervision to NeRF decoder | NeRF decoder · Image reconstruction in 3 camera views | rendering fidelity | Improved |
Closest papers by meaning
- GoT-R1: Unleashing Reasoning Capability of Autoregressive Visual Generation with Reinforcement Learning (ICLR 2026 · similarity 0.85)
- GSNeRF: Generalizable Semantic Neural Radiance Fields with Enhanced 3D Scene Understanding (CVPR 2024 · similarity 0.85)
- Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play (ICLR 2026 · similarity 0.85)
- PeRL: Permutation-Enhanced Reinforcement Learning for Interleaved Vision-Language Reasoning (NeurIPS 2025 · similarity 0.85)
- Spatial-SSRL: Enhancing Spatial Understanding via Self-Supervised Reinforcement Learning (CVPR 2026 · similarity 0.85)
- Fourier Guided Adaptive Adversarial Augmentation for Generalization in Visual Reinforcement Learning (AAAI 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
