Overview

Robot policy learning studies how robots can turn camera, depth, tactile, and proprioceptive inputs into actions for tasks such as manipulation, locomotion, and navigation. Common approaches include behavior cloning from demonstrations, reinforcement learning in simulation or on hardware, and diffusion, flow-matching, or vision-language-action policies that generate action sequences; world models and hierarchical planners can add foresight and long-horizon structure. Researchers evaluate these methods in simulation suites such as LIBERO, CALVIN, RoboTwin, and navigation benchmarks, as well as on physical robots, with growing attention to memory, contact, safety, and recovery from failure. A central tension is how to make policies generalize across tasks, robots, and changing conditions while remaining data-efficient, reactive, reliable, and fast enough to deploy.

Latest papers

405 papers42 papers in the last four weeks, down 58% on the four weeks before. 0.3% of all new papers.

Sep 28, 2026

WAM-OPD: Sharpening World Action Models via On-Policy Distillation

Pretrained world action models (WAMs) provide generalist capabilities across diverse robotic manipulation tasks, yet improving target-task performance to an expert level without degrading pretrained skills remains challenging. We explore on-policy distillation (OPD) for WAMs and introduce WAM-OPD. WAM-OPD inherits the advantage of OPD methods that transfer task-specific teacher knowledge under the student's own induced distribution, rather than directly fitting the student to a narrow task-specific data distribution. However, in closed-loop manipulation, the observation histories change as the student policy evolves, requiring fresh environment rollouts to remain on-policy. Applying OPD to WAMs entails repeated data collection, which is costly even in simulation and often impractical on real robots. To avoid repeated environment rollouts during distillation, we introduce prefix-weighted trajectory replay (PWTR). PWTR uses a fixed trajectory pool composed primarily of initial-student rollouts, supplemented with task-specific teacher rollouts to broaden trajectory coverage. For each trajectory replayed from this pool, PWTR conditions the current policy on successive stored histories to generate fresh denoising paths, along which the task-specific teacher provides supervision. Although these denoising paths are refreshed as the policy evolves, the replayed environment trajectories remain fixed. PWTR therefore reweights per-decision distillation losses using proxy importance weights derived from path scores accumulated over the trajectory prefix preceding each decision to mitigate the resulting shift in the history distribution. Simulated and real-world experiments demonstrate task adaptation without additional environment interaction during distillation. In both settings, WAM-OPD improves target-task performance while retaining near-initial performance on tasks excluded from adaptation.
Sep 28, 2026

FailPatch: Failure Residual Patching for Vision-Language-Action Models

Vision-Language-Action (VLA) policies are typically adapted using successful demonstrations, which provide direct action supervision but rarely cover failure-prone states. Deployment failures expose these states, yet lack the corrective actions needed for conventional supervised learning. We propose FailPatch, a failure-driven residual patching framework that decouples action supervision from execution-reliability supervision. Successful demonstrations ground how the policy should act, while deployment trajectories indicate when its behavior becomes unreliable. We further observe that action hidden representations exhibit clear linear separability between reliable and failure-associated states while directly conditioning action generation. Building on these insights, FailPatch introduces a Null-gated Residual Expert Bank into the action hidden space of a frozen VLA policy. A unified Preserve--Redirect--Trust objective retains the original policy in reliable states, selects residual experts in failure-associated states and redirects representations from failure regions toward success-associated regions under bounded intervention. With only 0.52% trainable parameters, FailPatch improves success rates by 11.0 percentage points on four long-horizon RoboTwin tasks under clean evaluation, 9.5 percentage points under clean-to-random generalization, and 16.7 percentage points over the baseline across three real-world tasks. Project and code: https://github.com/yupeng-2003/FailPatch.
Sep 27, 2026

Principal Steering Subspaces for Online Adaptation of Frozen Generative Robot Policies

Generative robot policies provide expressive behavior priors, but updating a large diffusion or flow-matching model through online interaction is costly. Latent-space reinforcement learning avoids updating the pretrained generator by controlling its initial sampling noise, yet high-dimensional noise can have strongly anisotropic effects on decoded actions. We introduce Principal Steering Subspaces (PSS), a forward-query interface that constructs a fixed low-dimensional control basis from finite-difference decoder responses. Soft Actor-Critic controls the leading response directions, while the orthogonal complement is independently resampled from the Gaussian prior at each query. On three RoboMimic tasks with diffusion and flow-matching policies, response spectra reveal substantial concentration. Across five matched task-generator pairs, the training curves indicate that PSS generally converges faster and exhibits more stable late-training behavior than full-latent control, while achieving stronger final performance overall. Controlled Diffusion-Square ablations further show that leading-response directions outperform random and least-responsive subspaces of equal dimension. We further integrate PSS with a frozen, closed-source 3B-parameter vision-language-action (VLA) policy in a humanoid learning system with synchronous transition collection, reset-time optimization, and latency-aware asynchronous deployment. In an exploratory screwdriver-placement evaluation, success is observed in 2/10 trials for the frozen VLA policy and 6/10 after SAC+PSS adaptation. These results support decoder-response geometry as a practical basis for online adaptation of frozen generative robot policies.
Sep 27, 2026

TimelyDAgger: Timing-Aware Expert Querying for VLA Policy Improvement

DAgger improves robot policies by aggregating expert supervision from states visited during policy execution. Robot-gated DAgger automates expert queries, allowing the robot to decide when to request expert takeover. While existing gates emphasize detecting the need for assistance, takeover timing also shapes the content of these demonstrations and their value for policy learning. We propose TimelyDAgger, combining Bridge-PCA monitoring of internal vision-language-action (VLA) features with Feedback-guided Threshold Adaptation based on expert behavior to improve takeover timing. We introduce an evaluation framework linking failure detection, takeover timing, and policy improvement, including Target-Aligned Supervision Ratio (TASR) for assessing supervision quality without retraining. Experiments show that takeover timing affects policy learning, with TimelyDAgger achieving competitive failure detection and higher post-training success in most evaluated settings under matched expert-action budgets. Project website: https://seen-e.github.io/TimelyDagger/.
Sep 27, 2026

Beyond State-as-Action: Exploiting Command-State Discrepancy for Robot Imitation Learning

Constructing action targets from measured robot motion is an established approach in imitation learning. Under interaction constraints, however, command-state discrepancy may reflect control demands that motion alone does not capture. We investigate when this information matters and how to exploit it. Across three real-robot tasks, task and phase analyses reveal larger supervision gaps under constrained interaction, while selective command retention provides evidence of locally useful command information. Building on these findings, we propose Command-State Discrepancy Weighting (CSDW), which accounts for robot response times and combines subsequent progress, persistent unmet demand, and demand changes into continuous weights for command supervision. The method requires no task-phase annotations or changes to policy architecture or inference. CSDW improves over uniform command supervision on constrained tasks, while methods perform similarly in the less constrained task. Project page: https://seen-e.github.io/CSDW/.
Sep 26, 2026

Copper-Policy: Focus on the Representation for Robust Robot Manipulation

World Action Models (WAMs) acquire behavioral priors by modeling future scene evolution, but predicting detailed futures in pixel or latent space incurs substantial cost. Recent evidence that co-training gains persist without test-time generation raises a question: what must a WAM learn to improve control? We introduce Copper-Policy, which learns a compact World representation with the policy rather than relying on a predefined target space. Through temporal joint-embedding prediction, it predicts future observation embeddings conditioned on task intention without reconstructing pixels. This prediction and action decoding shape the representation jointly, while the policy retains access to current-frame spatial detail for execution. Representation analyses show that the learned features better separate task-driven change from perturbations and provide complementary information for control. Compact prediction targets reduce training tokens per sample, enabling a 2B-parameter model trained in 9.67 hours on 8×\times RTX 5090 GPUs and 6×\times faster than Fast-WAM on matched A100 GPUs. Copper-Policy outperforms every compared method without embodied pretraining on RoboTwin and several embodied-pretrained VLAs on LIBERO-Plus (80.85%). On three challenging real-robot tasks, it performs comparably to π0.5π_{0.5} and attains a higher average score. Together, these results show that Copper-Policy combines strong control performance with efficient training.
Sep 26, 2026

RE-0: Verified Recursive Improvement of Embodied Code-as-Policy Agents through Local On-Policy Distillation

Code-as-Policy agents accomplish long-horizon embodied tasks by generating and executing code, yet continually improving them with teachers that are stronger but not globally reliable remains a key challenge. Existing distillation methods typically treat the teacher's complete behavior as the supervision target and thus misassign training credit on states where the teacher fails. We propose RE-0, a recursively verified policy improvement framework: rather than assuming that the teacher globally outperforms the student, RE-0 requests local corrections from the teacher on the student's own failure histories and checks in the environment whether each correction is genuinely beneficial; verified corrections yield immediate improvement. Building on this, we propose RE-OPD, which turns verified interventions into supervision for on-policy distillation. Only counterfactually verified teacher interventions provide distribution-level supervision, weighted by their measured local benefit, and the improvement they induce is projected back into the standalone student, so both where supervision is applied and how much credit the teacher receives co-evolve with the student policy. We further prove that the student's per-round gain is lower-bounded by its verified intervention gain up to verification and projection error terms. Experiments across multiple Code-as-Policy embodied tasks show that RE-0 improves both teacher-assisted execution and the standalone student, and generalizes to novel robots and scenes.
Sep 26, 2026

Learning Social Navigation from Internet Videos in the Policy State Space

Training robust social-navigation policies requires simulators with diverse scene layouts, terrain, and human motion, but constructing such environments and specifying pedestrian behavior is costly. We propose an efficient pipeline that converts ordinary monocular walking videos directly into closed-loop social-navigation training environments in the policy's state space. Our key observation is that local social navigation primarily depends on two types of information: where the robot can traverse and how nearby pedestrians move. We therefore represent the static scene as a metric traversability map, which can be rigidly transformed under counterfactual robot motion, while directly replaying the pedestrian trajectories recovered from the video over time. This abstraction allows us to define the forward dynamics directly in the policy's state space and efficiently simulate counterfactual robot states without reconstructing or rendering photorealistic observations. The resulting policy achieves 81.2% success in the independent Arena benchmark, compared with 75.0% for the strongest baseline, and succeeds in 19/20 real-robot trials without policy fine-tuning. Project page: https://jiaming.im/VideoSocNav
Sep 26, 2026

RoboFFT: Finetuning generative robot policy via online reinforcement learning with forward process

Generative models, such as diffusion and flow-based models, have shown strong promise for robot policy learning by capturing complex and multimodal action distributions from demonstrations. However, policies trained solely with imitation learning often suffer from imperfect demonstrations and distributional shifts, while further improvement typically requires additional expert data. Reinforcement learning offers a natural solution through environment interaction, but effectively finetuning generative robot policies remains challenging due to the intractability of likelihood estimation. In this work, we propose RoboFFT, a forward-process reinforcement learning framework for finetuning generative robot policies, which applies forward noising to sampled actions and uses the weighted score / flow matching loss to construct a surrogate policy ratio for PPO-style updates. We evaluate RoboFFT with popular generative robot policies on representative simulation benchmarks, including long-horizon planning and sparse reward settings. Extensive experiments and analysis demonstrate that RoboFFT consistently improves performance while achieving better stability and training efficiency. We further integrate RoboFFT into a real world RL framework and demonstrate its effectiveness in real world tasks. Project website: https://student-of-holmes.github.io/RoboFFT/.
Sep 25, 2026

End-to-end QP-based policies: A unified perspective on robust control and robot learning

We present an end-to-end QP-based policy framework that enables systematic policy construction with minimal domain-specific design, while preserving the transparency and interpretability of model-based control. The proposed policy representation supports both domain-randomized model-based auto-tuning, where policy parameters are optimized over distributions of disturbances and model variations, and black-box policy construction, where the problem is formulated in terms of a (possibly) unknown model without requiring explicit notions of states, inputs, or the underlying system dynamics. We establish connections to existing policy representations and control paradigms, including robust control, multilayer perceptrons, and robot learning, and interpret our end-to-end QP policies as a common abstraction of these approaches. We validate the resulting framework in both simulation and hardware, demonstrating a broad range of applications spanning robustness, automatic policy tuning, and control under unknown system dynamics.
Sep 25, 2026

Learning Robot Policies from Sparse Success Signals via STL-Guided Stein Variational Policy Gradient

Learning robot policies for tasks with sparse success signals is challenging when completion depends on coordinated actions, precise contact outcomes, or satisfying several conditions together. Intricate physical interactions with the world further complicate these requirements. Prior work using conventional reward shaping mechanisms provides dense feedback but local progress might not translate into eventual task completion. We present Signal Temporal Logic-guided Stein Variational Policy Gradient (STL-SVPG), a population-based method that uses smooth STL robustness as a trajectory-level training objective. Differentiating this objective through the dynamics assigns credit to policy actions according to their effect on the complete task specification, rather than local progress alone. We evaluate the approach on six quadcopter and manipulator tasks that involves event-triggered responses, strictly ordered behavior, responses within specified deadlines, and physical interaction with the world. STL-SVPG achieves the highest mean success rate among the compared methods on five of six benchmarks. Simulation-trained policies trained in simulation transfer temporal and contact task behavior to the real world.
Sep 25, 2026

Learning to Leverage Compliance: A Policy-Admittance Learning Framework for Robotic Insertion

Policy learning and compliant control offer a promising route to reliable autonomous assembly under pose errors and contact uncertainty. However, combining them does not ensure coordination: the policy may continue pushing against contact while the controller yields, producing sustained loading with limited progress. To address this problem, we propose LeCo (Leverage Compliance), a policy-admittance learning framework that guides a visual policy through execution-time interaction under fixed admittance. A multirate feedback mechanism aggregates high-rate contact-interaction records into policy-transition rewards. An integrated conflict cost then characterizes sustained policy-loading/controller-unloading opposition, while a directional high-force tail cost captures continued-loading events within a transition. Together with task completion, these costs encourage the policy to leverage compliance with less unproductive loading. We evaluate LeCo on four real connector-assembly tasks, obtaining an aggregate success rate of 94%. Across tasks, mean successful-trial resultant-force and torque peaks decrease by approximately 30% and 64% relative to the comparison baseline. Reward ablation further shows that adding conflict shaping reduces median successful-trial contact-conditioned conflict density by approximately 53%. These results support learning to leverage fixed compliance by turning multirate policy-admittance interaction into complementary reward signals for effective, lower-load insertion.
Sep 24, 2026

HuGo: LLMs as Whole-Body Policy Code Designers for Humanoid Loco-Manipulation

For humanoids to be useful in everyday environments, they must perform a wide range of tasks that couple locomotion and manipulation. Existing approaches commonly acquire a loco-manipulation policy through reward engineering or demonstrations followed by task-specific training, making it costly to scale to new tasks. In this work, we propose a hierarchical approach to humanoid loco-manipulation that eliminates these per-task requirements. HuGo, Humanoid policy code Generation, uses a Large Language Model (LLM) to generate executable, closed-loop high-level policy code from a task description on top of a frozen low-level whole-body policy. Given the task, observation, and command specifications, the LLM constructs the task logic in code. HuGo then refines the policy from its rollouts using numerical trajectories and selected video frames to produce feedback and targeted code updates. Across five simulation tasks, using two different low-level policies, HuGo substantially outperforms a high-level reinforcement learning baseline and approaches the performance of a demonstration-based baseline. We achieve this level of performance without task-specific reward design or demonstration collection. We further demonstrate zero-shot transfer of simulation-generated policies to hardware and show that applying the same refinement loop to real-world rollouts can further improve transfer performance without expert demonstrations or policy retraining. Project website is https://iconlab.negarmehr.com/HuGo/
Sep 24, 2026

Privacy-Preserving Prompted Policy Search for Robotic Control

Large language models (LLMs) have recently demonstrated promising capabilities as in-context policy optimizers for Reinforcement Learning (RL), enabling policy search driven by both numerical reward signals and natural language reasoning. However, deploying such methods in practice requires transmitting raw policy parameters and rewards history to cloud-based LLM APIs, exposing proprietary control strategies to third-party service providers. To address this issue, this paper introduces Privacy-Preserving Prompted Policy Search (PP-ProPS), a framework that enables LLM-guided policy optimization while keeping policy and environmental parameters confidential. PP-ProPS encodes policy parameters and reward values using secret client-side transformations before they are included in each API request, ensuring that the LLM provider observes only encoded policy parameters and scaled reward information. Furthermore, unlike Vanilla ProPS, the proposed framework does not require the true optimal episodic return to be known or disclosed to the LLM. Beyond protecting the optimization data, PP-ProPS improves the search process in two ways. First, it provides the LLM with individual reward components instead of only a single total return, offering more informative feedback about each candidate policy. Second, it uses a bounded history that prevents the prompt from growing indefinitely, improving search with high-dimensional policies and supporting the use of open-weight LLMs. The proposed PP-ProPS is evaluated on both continuous and discrete control problems spanning Multi-Joint dynamics with Contact (MuJoCo) locomotion, classic control, highway driving, and robotic arm manipulation. Compared to Vanilla ProPS, the proposed PP-ProPS outperforms ProPS in seven of the ten evaluated tasks, and surpasses conventional RL methods including PPO, SAC, and TRPO, in five of the six tasks.
Sep 24, 2026

Training-free Behavior Cloning

Neural behavior cloning compresses demonstrations into large models, making individual actions difficult to trace and policy updates costly. Retrieval policies retain access to demonstrations but struggle with mismatch between recorded and live behavior. We introduce Behavior Predictive Control (BPC), which synthesizes policies without end-to-end policy training by combining an action-aware retrieval metric, a Hankel-based action-continuation prior, and a closed-form one-step residual correction. Inspired by behavioral systems theory, BPC predicts future actions by blending stored observation-action data that best reconstructs the recent runtime observation--action history. Across simulated benchmarks and real-robot deployments, BPC is competitive with learned policies such as π0.5π_{0.5} (surpassing it in some cases), while reducing policy fitting from hours to seconds on consumer GPUs and supporting closed-loop control upwards of 75 Hz on a Jetson Orin Nano. The retrieved demonstration windows and their coefficients also provide an intrinsic estimate of task progress. Retaining demonstrations within the deployed policy makes its predictions traceable to supporting trajectories and enables behavior revision through the demonstration bank.
Sep 24, 2026

Self-Adaptive VLA for Robust Robot Deployment

While Vision-Language-Action (VLA) models demonstrate impressive capabilities in robotic manipulation, their memoryless nature renders them brittle to test-time environment shifts, particularly hardware shifts caused by wear or imperfect calibration. Enabling these models to self-adapt during deployment without requiring continuous on-site recalibration remains a critical bottleneck for real-world scalability. In this work, we introduce Self-Adaptive VLA, a novel post-training recipe that enables the policy to iteratively adapt to deployment-time hardware shifts leveraging its own rollouts as context. To do so, we first collect policy rollouts under deliberately injected hardware shifts. We then transform the base policy's training data into shift-conditioned expert demonstrations by pre-compensating the expert actions for these known shifts. Next, we introduce a lightweight, plug-in context encoder that compresses the context, including visual observation, proprioception, and actions in the shifted environment, into a latent context token. This token modulates the policy through adaptive layer normalization (AdaLN). Furthermore, we find that context tokens can be ensembled, allowing the policy to iteratively self-correct and mitigate failures step by step. Extensive experiments across four precision-critical bi-manual and dexterous manipulation tasks show that Self-Adaptive VLA recovers over 80% of the base policy's performance under hardware shifts, such as actuation bias and joint encoder offsets. Moreover, Self-Adaptive VLA enables more robust deployment to new workstations compared to the base policy. Our approach provides a pathway for robust large-scale real-world robot deployments and easier maintenance. See videos at https://icefoxzhx.github.io/self-adaptive-vla.
Sep 24, 2026

Beyond Spatial Benchmarks: From Spatial Reasoning to Navigation

Does progress on spatial reasoning benchmarks translate into better navigation? Existing benchmarks test isolated inferences from images or videos, with little connection to downstream navigation. Our analysis reveals a gap between benchmark-oriented spatial specialization and navigation performance, and shows how aligning spatial supervision with navigation goals, phases, and decision learning improves navigation. Guided by these findings, we build Spatial-Nav-100K and fine-tune in two stages, i.e. first learning a shared spatial-navigation foundation, and then specializing each phase with the abilities it relies on. We further introduce Spatial-NPD, where a teacher conditioned on spatial priors produces grounded action preferences and distills them into a student policy, so no explicit spatial reasoning is needed at inference. With 45 A100 GPU-hours of policy training, our 8B model reaches SR/SPL of 77.4/35.4 on HM3D-v0.2, 60.2/30.5 on HM3D-v0.1, and 47.9/20.6 on train-unseen MP3D. It outperforms several systems that rely on closed-source models or thousands of GPU-hours of training, at 148 ms per action step. All code and datasets will be publicly available at https://github.com/ylwhxht/Spatial-Nav.
Sep 24, 2026

RACaP: Agentic Reasoning, Acting, and Coding as Policies for Evolvable Robot Learning

General-purpose robot agents must learn from experience, transfer to new tasks, and act efficiently. Code as Policies (CaP) methods generate and repair programs at runtime, incurring latency and entangling reusable mechanisms with task-specific decisions. We introduce RACaP, an agentic framework that moves coding to evolution and uses a Reasoning-and-Acting (ReAct) loop to call frozen, typed Policy APIs at deployment. A two-phase strategy combines capability curriculum learning with autonomous self-evolution to improve the APIs, the ReAct harness, and experience memory. The APIs encode reusable physical mechanisms while exposing arguments for runtime adaptation. ReAct combines task-specific working memory, long-term experience memory, and visual feedback to select actions, verify outcomes, and recover from failures without modifying source code. RACaP achieves 54.4% success on LIBERO-90, 45.0% on zero-shot LIBERO-PRO, and 46.0% on LIBERO-Long, compared with at most 4.0% for CaP baselines on long-horizon tasks. On LIBERO-PRO, it achieves 2.5 times the success rate of CaP baselines and a 1.9-fold speedup in median policy time. For efficient on-robot deployment, rejection-sampled fine-tuning distills GPT-5.6 ReAct decisions into Qwen3-VL-8B-Instruct, yielding a 13.2-fold per-decision inference speedup and reducing repeated physical calls from 16 to 4. These results show that separating reusable code from runtime decisions supports continued evolution, effective transfer, and efficient long-horizon control.
Sep 24, 2026

EgoSpeedUp: Transferring Human Manipulation Tempo to Robot Policies

Robot manipulation policies trained through imitation learning inherit not only the demonstrated behavior but also the conservative execution tempo of robot demonstrations. Existing acceleration approaches can execute faster than the original demonstrations, but determine the appropriate acceleration primarily from robot-side information or a predefined set of tempo factors, leaving open how to obtain a task-appropriate reference for how fast each manipulation phase should progress. We introduce EgoSpeedUp, a framework that uses human manipulation as temporal supervision for robot imitation learning. Our key insight is that human demonstrations naturally reveal task-appropriate, phase-wise manipulation tempo. Given slow robot demonstrations and human demonstrations of the same task, EgoSpeedUp aligns corresponding manipulation phases, estimates their relative execution tempos from multiple human demonstrations, and transfers the resulting phase-wise tempo by retiming the robot demonstrations. The retimed demonstrations are then used for standard behavior cloning, allowing the robot to retain its executable manipulation behavior while learning to perform it at a human-informed tempo. Across two real-world manipulation tasks, EgoSpeedUp improves the task success rate by an average of 25 percentage points (pp) while reducing successful execution time by 36.5%. These results demonstrate that human manipulation tempo provides an effective temporal reference for learning faster and more reliable robot policies.
Sep 24, 2026

Learning from Mixed-Quality Deployment Experience for Robot Manipulation

Robot policies deployed in real environments naturally accumulate mixed-quality experience, including successful executions, partial progress, and failures. Although these rollouts provide valuable information for further learning, directly incorporating them into imitation learning may reinforce undesirable behaviors, while offline reinforcement learning often suffers from unreliable value estimation under sparse rewards and limited data coverage. We consider a practical post-deployment setting where learning relies only on naturally accumulated autonomous rollouts, without additional human corrections or exploratory interaction. To effectively exploit such experience, we propose Predictive Action Chunk Learning (PACL). PACL first learns a predictive chunk-level critic that evaluates temporally extended action sequences and augments temporal difference learning with future latent prediction, providing richer supervision for long-horizon value estimation. The learned critic then converts chunk-level Q-values into discrete quality conditions, which guide a diffusion actor to learn jointly from these mixed-quality experiences without treating all behaviors as equivalent supervision. At inference, the actor generates multiple action chunks and the critic selects the highest valued candidate. Experiments across simulated and real-world robot manipulation tasks show that PACL consistently improves the pretrained policy and outperforms strong imitation learning and offline reinforcement learning baselines.
Sep 23, 2026

Uncertainty-Gated Exploration Noise Suppresses Task Collapse in Online RL Fine-Tuning of a Flow-Matching Vision-Language-Action Policy

Online reinforcement learning fine-tuning of pretrained flow-matching vision-language-action (VLA) policies promises robots that keep learning after deployment, but continued updates often destroy competence on individual tasks while the aggregate still looks healthy. We study this failure mode, which we call task collapse, under a matched small-compute budget on LIBERO-10 with a 450M-parameter SmolVLA policy trained by PPO with stochastic (SDE) sampling. Three exploration-noise policies differ in one live variable: a fixed noise scale, a ReinFlow-style learned noise network, and an uncertainty-gated controller that redistributes exploration across task streams from task-agnostic novelty and competence signals, without task labels or episode boundaries. Under the pooled definition, fixed noise collapses tasks in two of three seeds and learned noise in every seed measured to iteration 200, while the controller collapses none in any of its three seeds. Measured parameter displacement shows the controller's action expert keeps changing, while its mean applied noise is close to the fixed scale in the available logs. The matched comparison supports the controller's effect on task preservation; the separate contributions of its adaptation across states and over time are not disentangled. A lower fixed scale slows the decline but does not stop it. No arm improves on the behavior-cloning baseline in this budget. Two properties of that regime are measured beside this result, not offered as its cause: following the reference recipe, training runs in bfloat16 with no fp32 master copy, under which 96.02% of the action expert's elements stay bit-identical across three consecutive iterations, and an fp32 master copy at the reference learning rate collapses both arms in a single-seed observation. We release tools measuring per-task collapse under four definitions, rescoring noise and instrument tares.
Sep 23, 2026

Behavior-Aligned Action Tokenization for Robot Policy Learning

Autoregressive robot policies learn continuous control by predicting discrete action tokens from observations. Different tasks often share local motions, yet behavioral correspondence across demonstrations receives limited explicit supervision in existing tokenizers. Motions with different timing can therefore lack a shared representation despite following similar patterns. We propose Behavior-Aligned Action Tokenization (BAAT), which uses soft dynamic time warping (Soft-DTW) to select corresponding action chunks and aligns their quantized coordinates jointly with reconstruction. This objective encourages similar motions across tasks to occupy nearby quantized representations while retaining executable action detail. A history-conditioned diffusion decoder reconstructs continuous action chunks from these tokens, and a downstream autoregressive policy learns to predict them. We evaluate BAAT on selected tasks from three simulation benchmarks and two real robot tasks. BAAT achieves a mean simulation success rate of approximately 45.2%, exceeding OAT by approximately 7.2 percentage points. In the controlled LIBERO-All alignment ablation, policy success rises from 70.2% to 79.0% while trajectory replay success decreases. These results support behavioral correspondence as supervision for organizing shared motion structure in action tokenizers and improving downstream robot policy learning.
Sep 22, 2026

HiRE: Hindsight Reward Editing for Policy Finetuning

Pre-trained robot policies always require finetuning to adapt to specific environments. Reinforcement Learning (RL) offers high performance potential because it improves action optimality rather than simply mimicking data. However, such potential depends heavily on reward quality. Sparse rewards lack process feedback, human-designed rewards are costly and biased, and semantic rewards from foundation representations are often not control-centric. We propose Hindsight Reward Editing (HiRE), a training-free framework to break this reward bottleneck. HiRE bridges the broad knowledge of foundation representation models with physical control awareness, by contrasting successful and failed trajectories in hindsight. It calibrates foundation representation models by identifying "trap states" that are predicted as high-rewarding states yet eventually result in failure, and vice versa. HiRE explicitly penalizes these traps while boosting rewards for critical successful states. This approach can be flexibly compatible with any foundation representations and RL algorithms. Experiments show that HiRE consistently outperforms other reward recipes by delivering dense, control-aware feedback that prevents value function collapse and reward hacking, thereby achieving superior sample efficiency, stable policy updates, and higher performance ceilings, e.g., at least 3x performance of the base policies. Qualitative results are at https://hire-project.github.io .
Sep 22, 2026

Generalists Act, Specialists Intervene: Modular Stage-Selective Reinforcement Learning for Vision-Language-Action Manipulation

Vision-language-action (VLA) models often struggle in the precision-critical phases of multi-stage manipulation tasks. To mitigate this issue, VLA models can be used in conjunction with reinforcement learning (RL) specialists that are specifically trained to handle the precision-critical phases. However, the coordination between the base VLA model and the RL specialists, which dictates when a specialist should take over from the base VLA and vice-versa, remains an open research question. In this paper, we address this gap by introducing RouteRLT, a modular framework that coordinates a generalist VLA, used as the default controller, with designated precision-critical RL specialists. At a high level, our framework trains a phase-aware coordination mechanism that handles handoffs between the generalist and the specialists. We evaluate RouteRLT on the LIBERO and LIBERO-Plus benchmarks, as well as on a physical connector pickup and insertion task. Overall, we find that RouteRLT improves success on LIBERO, and retains net gains on LIBERO-Plus. On the physical task, RouteRLT completes 65.7% of trials, compared with 8.6% for the baseline. Altogether, these results demonstrate that learned coordination builds on generalist VLA capabilities to improve task completion in precision-critical manipulation.
Sep 22, 2026

MAVP: Map-Aware Visuomotor Policies for Mobile Manipulation

Successful mobile manipulation requires coordinated base and arm motion while maintaining accurate spatial positioning. However, demonstration-trained policies can struggle to realise the intended base motion reliably, leading to spatial misalignment and subsequent manipulation failures. We present MAVP (Map-Aware Visuomotor Policies), a framework that improves execution reliability by predicting explicit base-pose targets and tracking them using localisation feedback. MAVP reconstructs a static map from teleoperated demonstrations and expresses demonstrated base trajectories in a shared map frame, providing consistent spatial supervision across demonstrations. At execution time, the policy receives RGB observations, joint states, and the robot's current map-frame base pose, and jointly predicts target base poses, arm actions, and gripper actions. A low-level controller tracks the predicted base targets using feedforward motion and pose error feedback, enabling correction of execution deviations. We additionally use pose-noise augmentation during training to improve robustness to errors in the policy's pose input. Across six real-world manipulation tasks and three policy families, MAVP achieves higher task success rates than unanchored velocity control in all tasks. Videos and additional results are available at https://123qwedsa123.github.io/mavp/.
Sep 20, 2026

BarrierFormer: Transformer-Guided Predictive Barrier Enforcement for Safe Robot Control

Control barrier functions (CBFs) have become one of the most popular tools for encoding and enforcing state constraints in safety-critical robotics. Standard CBF approaches are inherently myopic in nature as they enforce safety only at the current time step. Consequently, the system can be driven toward the boundary of the safe set where no feasible safe control exists at a future timestep. Model predictive control (MPC) based approaches address this by enforcing state constraints over a receding horizon. However, such approaches generally require the model to be known for solving a constrained optimization problem at every step, which is computationally expensive for real-time deployment. We propose BarrierFormer, a barrier-supervised transformer framework that addresses these limitations by encoding rollout-level CBF constraints in learning a model-free safe policy. A causal transformer encodes observation-action history, autoregressively generates a predictive rollout through the dynamics head to replace the model, and provides a residual correction to a nominal controller through the action head to replace the online computation. A barrier critic operating on local observations evaluates CBF constraint violations along this rollout, and a safety teacher computes barrier-consistent actions satisfying these constraints as direct supervision targets for the learned control policy. During inference, the policy maps observation-action history to control actions without any online optimization or model knowledge, enabling real-time model-free predictive safety enforcement. Evaluations across linear and nonlinear, 2D and 3D dynamical systems for safe goal-directed navigation demonstrate that BarrierFormer outperforms existing reinforcement learning (RL)-based, diffusion-based, MPC-based, and transformer-based approaches in safety rate and inference latency.
Sep 19, 2026

Stable and Efficient Real-World Online VLA Post-Training via Asynchronous Replay-Anchored Policy Improvement

Online post-training of vision-language-action (VLA) models requires efficient use of robot interaction and reliable policy improvement from continually collected experience. We propose asynchronous Replay-Anchored Policy improvement (RAPolicy), a framework that performs rollout and learning concurrently while grounding both critic and actor updates in replayed behavior. The critic learns chunk-level values from recorded actions and constructs Bellman targets without predicting next actions, reducing computation and dependence on action-value estimates outside replay coverage. The one-step flow actor reuses the initial noise stored during rollout and learns through advantage-weighted conditional likelihood, directly supervising the action mapping used for execution. We evaluate RAPolicy across four single-task settings and one joint five-task setting in the real world, with online training budgets of only 1--2 hours. Starting from policies fine-tuned on just 10 demonstrations per task, RAPolicy rapidly adapts to new single tasks and achieves an average 86.3% success rate. In the joint multi-task setting, RAPolicy improves overall success rate from 52% to 88% while preserving performance on already reliable tasks and improving weaker capabilities. Overall, RAPolicy substantially outperforms the baselines in aggregate task success while requiring fewer human interventions, demonstrating stable policy improvement and high online training efficiency. Project page: https://flyfaerss.github.io/RAPolicy.
Sep 18, 2026

SeeQ: Training Generalist Value Functions for Long-Horizon Robotic Manipulation

Despite rapid progress, generalist robot policies remain brittle on complex, long-horizon tasks that comprise multiple stages or require repeated attempts and deliberation on the same underlying stage before success. Q-value functions can improve these policies by ranking candidate actions or guiding policy improvement, but learning from sparse task-level rewards entails long credit-assignment horizons, difficult Bellman backups, and broad data-coverage requirements. We introduce SeeQ (Subtask-elicited Q-functions), which instead learns Q-values for the currently active subtask. This shortens the value-prediction horizon and enables effective learning with temporal-difference (TD) objectives. During training, subtask-level annotations present in offline robot data provide the decomposition and enable learning from broad, potentially suboptimal robot datasets. To eliminate the need for human annotations or modular subtask prediction systems at test time, our Q-function architecture is trained to autoregressively predict the active subtask in natural language before estimating its value. We instantiate SeeQ using a base vision-language backbone, pretrain it on diverse open-source robot manipulation data, and finetune it on downstream tasks. Across four real-world manipulation tasks on two bimanual robot platforms, the SeeQ value function substantially improves best-of-N policy steering.
Sep 18, 2026

CARF: Contrastive Attraction-Repulsion of Failure-Guided Flow Matching

Robot demonstration collection often produces imperfect or failed trajectories in addition to successful demonstrations. Existing methods typically exploit failed trajectories by identifying segments that still make progress toward task completion, but largely overlook failure-critical behaviors that directly lead to task failure. Here we argue that these two types of segments provide fundamentally asymmetric supervision: progressive segments should be imitated, whereas failure-critical segments should be explicitly avoided. Based on this observation, we propose CARF, a Contrastive Attraction-Repulsion of Failure-guided framework for learning from imperfect robot data. CARF introduces a progress-based importance scorer, trained solely on successful expert demonstrations and its perturbation results, to estimate step-wise contributions toward task completion and identify informative regions in failed trajectories. These scores guide a unified flow-matching objective that attracts the policy toward progressive behaviors and repels it from failure-critical ones, while excluding ambiguous segments. This enables more comprehensive utilization of imperfect data and avoids unreliable supervision from ambiguous failure segments. Extensive experiments in simulation and the real world demonstrate consistent improvements over competing baselines across diverse failure scenarios, with ablations further validating the effectiveness of the proposed scoring and attraction-repulsion mechanisms. Our website is https://zhao-sq.github.io/carf/#.
Sep 18, 2026

DPed-VLN: A Benchmark for Socially Compliant Vision-and-Language Navigation in Dynamic Pedestrian Environments

Vision-and-language navigation (VLN) has advanced rapidly in static indoor environments, but robots operating in human-populated spaces must ground language while responding to moving pedestrians and social-safety constraints. We present DPed-VLN, a Habitat 3.0 benchmark for dynamic-pedestrian VLN that couples 33,093 navigation episodes with paired global and prior-augmented instructions, ORCA-controlled humanoid pedestrians, socially constrained expert paths, and metrics that jointly assess navigation efficiency and social safety. DPed-VLN separates ordinary goal-oriented route guidance from prior-augmented instructions that expose dynamic-pedestrian cues for controlled analysis. To instantiate the benchmark, we introduce DPet (Dynamic Pedestrian-aware Network), a pedestrian-aware policy network trained with reinforcement learning and imitation learning. We further adapt representative state-of-the-art VLM-based navigation models, including NaVILA and StreamVLN, to DPed-VLN through LoRA fine-tuning. Experiments show that LoRA adaptation improves zero-shot VLM baselines in several success and safety metrics, especially reducing StreamVLN's collision rate. Among the evaluated methods, DPet-RL achieves the highest SR, SPL, and STL.