Research map / Reinforcement learning / Visual and representation RL
Self-Improving Loops for Visual Robotic Planning
Calvin Luo, Zilai Zeng, Mingxi Jia, Yilun Du, Chen Sun
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 |
|---|---|---|---|
| use visual planning (SILVR) instead of behavior-cloning improvement loop | SILVR vs BCIL · MetaWorld 12 unseen tasks | success rate (%): BCIL 23.2 -> SILVR 44.2 (iter 4) | Improved |
| self-improve with RL finetuning (DSRL) instead of SILVR | DSRL · MetaWorld 12 unseen tasks | success rate (%): DSRL 9.4 -> 7.7 | Hurt |
| replace ground-truth success filter with VLM filter (ablation) | SILVR · MetaWorld 12 unseen tasks | success rate (%): 44.2 -> 38.4 (iter 4) | Mixed |
| finetune both video model and IDM during self-improvement (ablation) | SILVR · MetaWorld | success rate (%): 26.8 / 29.8 -> 44.2 (iter 4) | Improved |
| distill final video planner into diffusion policy | SILVR · MetaWorld | success rate (%): 44.2 -> 49.2 | Improved |
| add internet video prior (AnimateDiff) via IPA (ablation) | SILVR · Panda arm real-world tasks | success rate (%) | Improved |
| add internet video prior via IPA (ablation) | SILVR · MetaWorld | success rate (%) | No clear effect |
| skip data filtering (ablation) | SILVR · MetaWorld | success rate (%) | Hurt |
| skip data filtering (with internet prior) (ablation) | SILVR · Panda arm | success rate (%) | No clear effect |
| initialize with suboptimal (70% random action) demonstrations (ablation) | SILVR · MetaWorld 12 unseen tasks | success rate (%): 44.2 (expert) vs 41.0 (suboptimal) at iter 4 | Mixed |
| train DSRL with 60000 instead of 150 updates (ablation) | DSRL · MetaWorld | success rate (%): 8.3 vs 7.7 (iter 4) | No clear effect |
| run more SILVR iterations beyond 5 (ablation) | SILVR · MetaWorld | success rate (%): 51.1 (iter 5) -> 53.1 (iter 9) | No clear effect |
Closest papers by meaning
- GEVRM: Goal-Expressive Video Generation Model For Robust Visual Manipulation (ICLR 2025 · similarity 0.90)
- FLIP: Flow-Centric Generative Planning as General-Purpose Manipulation World Model (ICLR 2025 · similarity 0.90)
- Zero-Shot Robotic Manipulation with Pre-Trained Image-Editing Diffusion Models (ICLR 2024 · similarity 0.89)
- Video Language Planning (ICLR 2024 · similarity 0.89)
- Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulation (ICLR 2024 · similarity 0.88)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.88)
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.
