Research map / Reinforcement learning / Policy gradient and actor-critic
RFS: Reinforcement learning with Residual flow steering for dexterous manipulation
Entong Su, Tyler Westenbroek, Anusha Nagabandi, Abhishek Gupta
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 |
|---|---|---|---|
| add residual action steering to latent-noise steering (ablation) | DSRL · Simulated tasks average | success rate: 0.483 -> 0.861 | Improved |
| add residual action steering to latent-noise steering (ablation) | DSRL · Simulated stacking | success rate: 0.135 -> 0.951 | Improved |
| add latent-noise steering to residual RL (ablation) | ResiP · Simulated tasks average | success rate: 0.433 -> 0.861 | Improved |
| add latent-noise steering to residual RL (ablation) | Policy Decorator · Simulated tasks average | success rate: 0.286 -> 0.861 | Improved |
| restrict RL to initial noise and residual instead of full denoising trajectory | ReinFlow · Simulated tasks average | success rate: 0.409 -> 0.861 | Improved |
| RL fine-tune the flow-matching base policy with RFS | Flow Matching base policy · Simulated tasks average | success rate: 0.250 -> 0.861 | Improved |
| add RFS fine-tuning | Flow Matching base policy · Simulated pouring | success rate: 0.15 -> 0.873 | Improved |
| condition critic on executed action a_b+a_r instead of residual-only or concatenated actions (ablation) | offline RFS · Real-world grasping and pick-and-place | success rate | Improved |
| use offline RL fine-tuning instead of co-training | distilled sim policy · Real-world unseen objects | success rate | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReinFlow — ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning (NeurIPS 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation (NeurIPS 2025 · similarity 0.89)
- RLIF: Interactive Imitation Learning as Reinforcement Learning (ICLR 2024 · similarity 0.88)
- Implicit Drifting Policy: One-Step Action Generation via Conditional Expert Geometry (NeurIPS 2026 · similarity 0.88)
- FlowPolicy: Enabling Fast and Robust 3D Flow-Based Policy via Consistency Flow Matching for Robot Manipulation (AAAI 2025 · similarity 0.88)
- Improving Generative Behavior Cloning via Self-Guidance and Adaptive Chunking (NeurIPS 2025 · 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.
