Pith. sign in

REVIEW 46 cited by

ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.05450 v2 pith:VZ4ED6DW submitted 2025-02-08 cs.RO cs.AI

ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy

classification cs.RO cs.AI
keywords fine-tuningmodelsconrftonlinepolicyreal-worldapproachconsistency
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Vision-Language-Action (VLA) models have shown substantial potential in real-world robotic manipulation. However, fine-tuning these models through supervised learning struggles to achieve robust performance due to limited, inconsistent demonstrations, especially in contact-rich environments. In this paper, we propose a reinforced fine-tuning approach for VLA models, named ConRFT, which consists of offline and online fine-tuning with a unified consistency-based training objective, to address these challenges. In the offline stage, our method integrates behavior cloning and Q-learning to effectively extract policy from a small set of demonstrations and stabilize value estimating. In the online stage, the VLA model is further fine-tuned via consistency policy, with human interventions to ensure safe exploration and high sample efficiency. We evaluate our approach on eight diverse real-world manipulation tasks. It achieves an average success rate of 96.3% within 45-90 minutes of online fine-tuning, outperforming prior supervised methods with a 144% improvement in success rate and 1.9x shorter episode length. This work highlights the potential of integrating reinforcement learning to enhance the performance of VLA models for real-world robotic applications. Videos and code are available at our project website https://cccedric.github.io/conrft/.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 46 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Adapting Generalist Robot Policies with Semantic Reinforcement Learning

    cs.RO 2026-06 unverdicted novelty 7.0

    SARL optimizes language prompt inputs to generalist vision-language-action policies through online RL to solve complex long-horizon tasks by composing existing skills.

  2. SOLE-R1: Video-Language Reasoning as the Sole Reward for On-Robot Reinforcement Learning

    cs.RO 2026-03 conditional novelty 6.5

    A video-language model with per-timestep spatiotemporal CoT and dense progress prediction can serve as the sole reward for zero-shot online robot RL on 24 unseen manipulation tasks.

  3. X-NavDP: Generalizing Navigation Diffusion Policy to Novel Behavior and Embodiments with Group Q-score Reweighted Matching

    cs.RO 2026-07 conditional novelty 6.0

    Group Q-score reweighted matching plus self-bootstrapped trajectory perturbation lifts cross-embodiment diffusion navigation from 61% to 84% success in sim and ~10% to 65% in hard real cases.

  4. Do You Really Need to Pretrain Q-Functions for Online RL Fine-Tuning?

    cs.LG 2026-07 conditional novelty 6.0

    Naive Q-pretraining rarely helps online RL fine-tuning of a pretrained policy; IPE’s multi-policy rollouts close the Qπbase vs Qπ*RL gap and raise success ~1.26×.

  5. Look Before You Leap: Distilling Tree Search into Action Evaluation for Frozen VLA Models

    cs.RO 2026-07 conditional novelty 6.0

    Distilling MCTS returns into a lightweight Q-model lets frozen VLAs select higher-return actions at test time, lifting success rates and letting a 9B model beat a 27B baseline at lower latency.

  6. One Demonstration Is Enough for Real-World Robotic Reinforcement Learning

    cs.RO 2026-07 unverdicted novelty 6.0

    AutoSERL achieves strong performance on six real-world robot manipulation tasks using RL guided by a single demonstration via sliding-window intervention, safety recovery, and automatic termination.

  7. Improving Vision-Language-Action Model Fine-Tuning with Structured Stage and Keyframe Supervision

    cs.RO 2026-06 unverdicted novelty 6.0

    StaKe adds lightweight auxiliary heads for manipulation stage identification and next-gripper-transition keyframe prediction to VLA fine-tuning, reporting relative success rate gains of 14% in bimanual simulation and ...

  8. Learning Process Rewards via Success Visitation Matching for Efficient RL

    cs.LG 2026-06 unverdicted novelty 6.0

    Success Visitation Matching uses a discriminator to turn sparse outcome rewards into dense process rewards by matching visitations of successful episodes, provably preserving the optimal policy and speeding up robotic...

  9. SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation

    cs.RO 2026-06 unverdicted novelty 6.0

    SARM2 presents RM, a multi-task stage-aware reward model achieving 80% lower value-estimation MSE, which when used in SPIRAL boosts manipulation task success from ~50% to near-perfect on several benchmarks.

  10. FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning

    cs.CV 2026-06 unverdicted novelty 6.0

    FiberTune is a new fine-tuning objective that preserves action-fiber visual residuals in VLA policies, yielding performance gains on simulation and physical robot tasks.

  11. Preference-Calibrated Human-in-the-Loop Reinforcement Learning for Robotic Manipulation

    cs.RO 2026-06 unverdicted novelty 6.0

    PACT calibrates credit assignment in HIL-RL by penalizing Bellman targets on suboptimal segments using counterfactual advantages from human-policy preference pairs, yielding 24.5% higher success rates and 1.3x faster ...

  12. Preference-Calibrated Human-in-the-Loop Reinforcement Learning for Robotic Manipulation

    cs.RO 2026-06 conditional novelty 6.0

    PACT uses demo-trained progress localization plus intervention preference pairs to correct inflated Bellman targets and align the actor, raising average real-robot success by 24.5% over HIL-SERL.

  13. Hand-in-the-Loop: Improving VLA Policies for Dexterous Manipulation via Seamless Hand-Arm Intervention

    cs.RO 2026-05 unverdicted novelty 6.0

    HandITL blends human intent with policy execution to eliminate gesture jumps in dexterous VLA interventions, cutting jitter by 99.8%, grasp failures by 87.5%, and yielding 19% better refined policies.

  14. Hand-in-the-Loop: Improving VLA Policies for Dexterous Manipulation via Seamless Hand-Arm Intervention

    cs.RO 2026-05 unverdicted novelty 6.0

    HandITL enables seamless human intervention in VLA policies for bimanual dexterous manipulation, cutting jitter by 99.8% and improving refined policies by 19% over standard teleoperation.

  15. UniSteer: Unified Noise Steering for Efficient Human-Guided VLA Adaptation

    cs.RO 2026-05 conditional novelty 6.0

    By inverting a frozen flow-matching robot policy with fixed-point iteration, UniSteer converts human corrections into noise-space supervision and lifts real-world VLA adaptation success from 20% to 90% in about 66 min...

  16. UniSteer: Unified Noise Steering for Efficient Human-Guided VLA Adaptation

    cs.RO 2026-05 unverdicted novelty 6.0

    UniSteer unifies human corrective actions and noise-space RL for VLA adaptation by inverting actions to noise targets, raising success rates from 20% to 90% in 66 minutes across four real-world manipulation tasks.

  17. Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs

    cs.RO 2026-05 unverdicted novelty 6.0

    A retrieve-then-steer method stores successful robot actions in memory and uses them to steer a frozen VLA's flow-matching sampler for better test-time reliability without parameter updates.

  18. Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs

    cs.RO 2026-05 unverdicted novelty 6.0

    Retrieve-then-steer stores successful observation-action segments in memory, retrieves relevant chunks, filters them, and uses an elite prior with confidence-adaptive guidance to steer a flow-matching action sampler f...

  19. Escaping the Diversity Trap in Robotic Manipulation via Anchor-Centric Adaptation

    cs.RO 2026-05 unverdicted novelty 6.0

    Anchor-Centric Adaptation escapes the diversity trap by prioritizing repeated demonstrations at core anchors over broad coverage, yielding higher success rates under fixed data budgets in robotic manipulation.

  20. Learning While Deploying: Fleet-Scale Reinforcement Learning for Generalist Robot Policies

    cs.RO 2026-05 unverdicted novelty 6.0

    Fleet-scale RL framework improves a single generalist VLA policy from deployment data to 95% average success on eight real-world manipulation tasks with 16 dual-arm robots.

  21. Learning While Deploying: Fleet-Scale Reinforcement Learning for Generalist Robot Policies

    cs.RO 2026-05 unverdicted novelty 6.0

    LWD is a fleet-scale offline-to-online RL framework that continually improves pretrained VLA policies using autonomous rollouts and human interventions, reaching 95% average success on real-world manipulation tasks.

  22. LaST-R1: Reinforcing Robotic Manipulation via Adaptive Physical Latent Reasoning

    cs.RO 2026-04 unverdicted novelty 6.0

    LaST-R1 introduces a RL post-training method called LAPO that optimizes latent Chain-of-Thought reasoning in vision-language-action models, yielding 99.9% success on LIBERO and up to 22.5% real-world gains.

  23. LaST-R1: Reinforcing Robotic Manipulation via Adaptive Physical Latent Reasoning

    cs.RO 2026-04 unverdicted novelty 6.0

    LaST-R1 reaches 99.8% average success on the LIBERO benchmark using one-shot warm-up plus LAPO reinforcement learning on latent physical reasoning, with up to 44% real-world gains on complex single- and dual-arm tasks.

  24. RL Token: Bootstrapping Online RL with Vision-Language-Action Models

    cs.LG 2026-04 unverdicted novelty 6.0

    RL Token enables sample-efficient online RL fine-tuning of large VLAs, delivering up to 3x speed gains and higher success rates on real-robot manipulation tasks within minutes to hours.

  25. MoRI: Mixture of RL and IL Experts for Long-Horizon Manipulation Tasks

    cs.RO 2026-04 unverdicted novelty 6.0

    MoRI dynamically mixes RL and IL experts with variance-based switching and IL regularization to reach 97.5% success in four real-world robotic tasks while cutting human intervention by 85.8%.

  26. Towards Long-Lived Robots: Continual Learning VLA Models via Reinforcement Fine-Tuning

    cs.RO 2026-02 unverdicted novelty 6.0

    LifeLong-RFT applies chunking-level on-policy reinforcement learning with Quantized Action Consistency Reward, Continuous Trajectory Alignment Reward, and Format Compliance Reward to fine-tune VLA models, achieving a ...

  27. TwinRL: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulation

    cs.RO 2026-02 unverdicted novelty 6.0

    TwinRL expands RL exploration via digital twin reconstruction and twin RL warm-up to guide real-world learning, reaching near-100% success with 20 minutes of on-robot time across four tasks.

  28. $\pi^{*}_{0.6}$: a VLA That Learns From Experience

    cs.LG 2025-11 unverdicted novelty 6.0

    RECAP enables a generalist VLA to self-improve via advantage-conditioned RL on mixed real-world data, more than doubling throughput and halving failure rates on hard manipulation tasks.

  29. PIGEON: VLM-Driven Object Navigation via Points of Interest Selection

    cs.RO 2025-11 conditional novelty 6.0

    PIGEON selects sparse Points of Interest for VLM-driven object navigation, achieving state-of-the-art zero-shot ObjectNav results in Habitat and improving a local 7B VLM via distance-based RLVR.

  30. DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models

    cs.LG 2025-10 unverdicted novelty 6.0

    DeepThinkVLA shows CoT improves VLA models only under decoding and causal alignment, delivering 97% success on LIBERO and 21.7-point gains via hybrid attention and SFT-RL training.

  31. SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning

    cs.RO 2025-09 conditional novelty 6.0

    SimpleVLA-RL applies tailored reinforcement learning to VLA models, reaching SoTA on LIBERO, outperforming π₀ on RoboTwin, and surpassing SFT in real-world tasks while reducing data needs and identifying a 'pushcut' p...

  32. Embodied-R1: Reinforced Embodied Reasoning for General Robotic Manipulation

    cs.RO 2025-08 conditional novelty 6.0

    Embodied-R1 uses a pointing-centric representation and reinforced fine-tuning on a 200K dataset to achieve state-of-the-art results on embodied benchmarks plus 56.2% success in SIMPLEREnv and 87.5% on real XArm tasks ...

  33. SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics

    cs.LG 2025-06 unverdicted novelty 6.0

    SmolVLA is a small efficient VLA model that achieves performance comparable to 10x larger models while training on one GPU and deploying on consumer hardware via community data and chunked asynchronous action prediction.

  34. Vision Language Action (VLA) Models for Unmanned Aerial Robotics and Bimanual Manipulation: A Review

    cs.RO 2026-07 accept novelty 5.5

    Bimanual VLA coordination strategies, training recipes, and continuous action chunking transfer to unmanned aerial systems; the survey maps 183 works and lists fourteen shared research directions.

  35. WorldSample: Closed-loop Real-robot RL with World Modelling

    cs.RO 2026-07 unverdicted novelty 5.0

    WorldSample generates synthetic transitions from a post-trained world model grounded in real rollouts and uses Policy-Paced Learning to improve RL policies, reporting 28% higher success rates and 59% fewer training st...

  36. AllDayNav: Lifelong Navigation via Real-World Reinforcement Learning

    cs.RO 2026-06 unverdicted novelty 5.0

    AllDayNav encodes scene dynamics into a large model's parameters via RL and a multimodal memory, achieving near-100% success rates in lifelong navigation and outperforming map-based and VLM baselines.

  37. DexPIE: Stable Dexterous Policy Improvement from Real-World Experience

    cs.RO 2026-06 unverdicted novelty 5.0

    DexPIE improves dexterous manipulation success rates by 37% over demo policies via real-world experience collection with adapted intervention, multi-stage DAgger, asynchronous relative-action inference, and optimality...

  38. BORA: Bridging Offline Reinforcement Learning and Online Residual Adaptation for Real-World Dexterous VLA Models

    cs.RO 2026-05 unverdicted novelty 5.0

    BORA combines offline RL critic training with online chunk-wise residual adaptation to raise average success rates of real-world dexterous VLA policies by 33% and up to 43% on unseen objects across five tasks.

  39. DyGRO-VLA: Cross-Task Scaling of Vision-Language-Action Models via Dynamic Grouped Residual Optimization

    cs.RO 2026-05 unverdicted novelty 5.0

    DyGRO-VLA is a two-stage optimization framework for cross-task scaling of Vision-Language-Action models via dynamic grouped residual optimization in RL.

  40. ALOE: Action-Level Off-Policy Evaluation for Vision-Language-Action Model Post-Training

    cs.RO 2026-02 conditional novelty 5.0

    ALOE uses chunked TD bootstrapping with a pessimistic Q-ensemble to enable action-level off-policy value estimation for advantage-weighted post-training of flow-based VLA policies, reporting consistent success-rate ga...

  41. VGAS: Value-Guided Action-Chunk Selection for Few-Shot Vision-Language-Action Adaptation

    cs.AI 2026-02 unverdicted novelty 5.0

    VGAS uses best-of-N selection with a geometrically grounded critic and explicit regularization to improve success rates of few-shot VLA policies under limited data and distribution shifts.

  42. Reflection-Based Task Adaptation for Self-Improving VLA

    cs.RO 2025-10 unverdicted novelty 5.0

    Reflective Self-Adaptation combines failure-reflective reinforcement learning with success-guided imitation learning to enable faster and more reliable task adaptation for pre-trained Vision-Language-Action models.

  43. Large VLM-based Vision-Language-Action Models for Robotic Manipulation: A Survey

    cs.RO 2025-08 unverdicted novelty 5.0

    This survey organizes large VLM-based VLA models for robotic manipulation into monolithic and hierarchical paradigms, reviews their integrations and datasets, and outlines future directions.

  44. Towards Precise Intent-Aligned VLA Aerial Navigation via Expert-Guided GRPO

    cs.RO 2026-06 unverdicted novelty 4.0

    EG-GRPO augments VLA aerial navigation with expert-guided group relative policy optimization and a faster simulation pipeline, claiming 2.13x success rate and 60.9% better intent alignment versus SFT baseline.

  45. EXPO-FT: Sample-Efficient Reinforcement Learning Finetuning for Vision-Language-Action Models

    cs.RO 2026-05 unverdicted novelty 4.0

    EXPO-FT enables pretrained VLA policies to reach 30/30 success on complex manipulation tasks using an average of 19.1 minutes of online robot data while outperforming prior RL approaches.

  46. A Survey of Reinforcement Learning for Large Reasoning Models

    cs.CL 2025-09 accept novelty 3.0

    A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.