Research map / Reinforcement learning / Visual and representation RL
VRL3: A Data-Driven Framework for Visual Deep Reinforcement Learning
Che Wang, Xufang Luo, Keith W. Ross, Dongsheng Li
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| remove stage 1 pretraining (S123 -> S23) (ablation) | VRL3 · Adroit | success rate | Improved |
| remove stage 2 offline RL (S123 -> S13) (ablation) | VRL3 · Adroit | success rate | Improved |
| use BC instead of conservative offline RL in stage 2 (ablation) | VRL3 · Adroit | success rate | Hurt |
| use naive RL updates in stage 2 (ablation) | VRL3 · Adroit | success rate and Q value | Hurt |
| remove safe Q target (ablation) | VRL3 · Adroit | success rate | Hurt |
| use conservative updates in stage 3 (ablation) | VRL3 · Adroit | success rate | Hurt |
| reduce encoder learning rate (ablation) | VRL3 · Adroit Relocate | success rate | Improved |
| freeze the encoder (learning rate scale 0) (ablation) | VRL3 · Adroit | success rate | Hurt |
| disable data augmentation (ablation) | VRL3 · Adroit | success rate | Hurt |
| contrastive learning in stage 2 on top of VRL3 | VRL3+FERM · Adroit | success rate | No clear effect |
| add stage 1 pretraining to FERM | FERM(VRL3) · Adroit | success rate | Improved |
| vary BC updates, CQL random actions and action std in stage 2 (ablation) | VRL3 · Adroit | success rate | No clear effect |
Closest papers by meaning
- VLMs-Guided Representation Distillation for Efficient Vision-Based Reinforcement Learning (CVPR 2025 · similarity 0.90)
- The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.88)
- COVR: Collaborative Optimization of VLMs and RL Agent for Visual-Based Control (AAAI 2026 · similarity 0.88)
- DMC-VB: A Benchmark for Representation Learning for Control with Visual Distractors (NeurIPS 2024 · similarity 0.88)
- When Should We Prefer State-to-Visual DAgger over Visual Reinforcement Learning? (AAAI 2025 · similarity 0.87)
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.
