mlsanityData to 7 Oct 2026

Revision, Not Restart: Revisable Visual Plans for Closed-Loop World-Action Models

arXiv 2609.35439 ↗First posted 28 Sep 2026cs.RO cs.CVcited by 0

On the map: World models → Embodied and robot world models

Abstract (arXiv)

World-action models use predicted visual futures to condition robot actions, yet execution feedback can invalidate parts of a prediction while leaving its task structure useful. We propose Revisable Temporal Planning (RTP), which maintains the visual future as a persistent action condition and revises it after feedback. Its central mechanism is a learned revision bridge: it resumes an intermediate state saved during visual generation and adapts its continuation to current observations. Visual and action supervision connect this revision to subsequent control. Time-aware history supplies observed evidence, and an adaptive policy selects retention, bridge revision, or fresh replanning from new noise before decoding the next action. On RoboMME and RMBench, RTP achieves task-averaged success rates of 48.6% and 84.8%, respectively. Matched comparisons support learned continuation; estimated checkpoint-source and action-prefix effects are positive but less precisely resolved. These results connect feedback-driven visual-plan revision to closed-loop task performance. Project Page: https://PLACEHOLDER.github.io/RTP/

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Embodied and robot world models, one of 11 directions in World models. Builds on 46 of its 50 references that are in the map. First posted on arXiv 28 Sep 2026.

Builds on

46 of 50 references.

Closest ancestors · 20

Same direction, newest first

  1. 2026Keep the Future, Drop the Rollout: RIFT for World Action Models12 Aug
  2. 2026Faster-WAM: Efficient Inference-Time Future Conditioning for Robust World Action Models5 Aug
  3. 2026Faster-WAM: Do World Action Models Need Deep Action Modules?3 Aug
  4. 2026SelfWAM: A Self-Grounded Unified World Action Model for Fast Robot Control1 Aug
  5. 2026World Action Planner: Generalizable Decision-Making with Action-Conditioned World Models30 Jul
  6. 2026FlowWAM: Optical Flow as a Unified Action Representation for World Action Models14 Jul
  7. 2026MemoryWAM: Efficient World Action Modeling with Persistent Memory18 Jun
  8. 2026LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot Policies14 Jun
  9. 2026AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing8 Jun
  10. 2026Light-WAM: Efficient World Action Models with State-Fusion Action Decoding6 Jun
  11. 2026Feedback World Model Enables Precise Guidance of Diffusion Policy15 May
  12. 2026OA-WAM: Object-Addressable World Action Model for Robust Robot Manipulation7 May
8 more
  1. 2026When to Trust Imagination: Adaptive Action Execution for World Action Models7 May
  2. 2026Unified 4D World Action Modeling from Video Priors with Asynchronous Denoising29 Apr
  3. 2026Action Images: End-to-End Policy Learning via Multiview Video Generation7 Apr
  4. 2026Fast-WAM: Do World Action Models Need Test-time Future Imagination?17 Mar
  5. 2026World Action Models are Zero-shot Policies17 Feb
  6. 2026Causal World Modeling for Robot Control29 Jan
  7. 2025Motus: A Unified Latent Action World Model15 Dec
  8. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusion7 Mar

Other references · 26

  1. 2022Scalable Diffusion Models with Transformerscited by 5,931
  2. 2022Flow Matching for Generative Modelingcited by 5,426
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  4. 2021SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equationscited by 1,903
  5. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,653
  6. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
  7. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  8. 2025RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulationcited by 477
  9. 2025X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Modelcited by 285
  10. 2025MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulationcited by 183
  11. 2025MemER: Scaling Up Memory for Robot Control via Experience Retrievalcited by 64
  12. 2025SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulationcited by 58
14 more
  1. 2026MEM: Multi-Scale Embodied Memory for Vision Language Action Modelscited by 55
  2. 2024Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Modelscited by 51
  3. 2026RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Designcited by 40
  4. 2026RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policiescited by 39
  5. 2026Native Video-Action Pretraining for Generalizable Robot Controlcited by 30
  6. 2026WorldKV: Efficient World Memory with World Retrieval and Compressioncited by 14
  7. 2026Addressable Memory for Video World Modelscited by 11
  8. 2026FutureRTC: Real-Time Robot Execution with Anticipatory-Conditioned Action Chunkingcited by 8
  9. 2025CycleManip: Enabling Cyclic Task Manipulation via Effective Historical Perception and Understandingcited by 7
  10. 2026FadeMem: Distance-Aware Memory Consolidation for Autoregressive Video Diffusioncited by 5
  11. 2026Chameleon: Control-Indexed Prospective Memory for Visuomotor Manipulationcited by 4
  12. 2026OVO-S-Bench: A Hierarchical Benchmark for Streaming Spatial Intelligence in Multimodal LLMscited by 4
  13. 2026Diffusion ReRoll: Revisable Denoising for Robotic Sequential Predictioncited by 4
  14. 2026TRACE: Temporal Retrieval with Anchored and Convergent Evidence for Long-Horizon Video Understandingcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026RecastVLA: From Past Interaction to Future Control with Adaptive Policy States ↗
  2. 2026Memory as Plans: World-Action Modeling with Memory-Grounded Planning
  3. 2025ViReSkill: Vision-Grounded Replanning with Skill Memory for LLM-Based Planning in Lifelong Robot Learning
  4. 2026Achieve What You Imagined: Learning to Align Actions with Visual Plans
  5. 2026Is the Future Compatible? Diagnosing Dynamic Consistency in World Action Models
  6. 2026An Action Is Worth One Patch: Unified World-Action Modeling with PatchWAM
  7. 2026RIFAR: Reliability and Forgetting-Aware Replay for Continual Robot Learning
  8. 2026Back to the Familiar Future: Failure Recovery for VLA Policies via Pre-Imagined Milestone Selection