mlsanityData to 7 Oct 2026

Reflective Planning: Vision-Language Models for Multi-Stage Long-Horizon Robotic Manipulation

arXiv 2502.16707 ↗First posted 23 Feb 2025cs.RO cs.AI cs.LGcited by 40

On the map: Multimodal LLMs / vision-language models → Embodied and 3D vision-language models

Abstract (arXiv)

Solving complex long-horizon robotic manipulation problems requires sophisticated high-level planning capabilities, the ability to reason about the physical world, and reactively choose appropriate motor skills. Vision-language models (VLMs) pretrained on Internet data could in principle offer a framework for tackling such problems. However, in their current form, VLMs lack both the nuanced understanding of intricate physics required for robotic manipulation and the ability to reason over long horizons to address error compounding issues. In this paper, we introduce a novel test-time computation framework that enhances VLMs' physical reasoning capabilities for multi-stage manipulation tasks. At its core, our approach iteratively improves a pretrained VLM with a "reflection" mechanism - it uses a generative model to imagine future world states, leverages these predictions to guide action selection, and critically reflects on potential suboptimalities to refine its reasoning. Experimental results demonstrate that our method significantly outperforms several state-of-the-art commercial VLMs as well as other post-training approaches such as Monte Carlo Tree Search (MCTS). Videos are available at https://reflect-vlm.github.io.

Key facts. Cited by 40 papers in the map (data through 7 Oct 2026). Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 33 of its 47 references that are in the map. First posted on arXiv 23 Feb 2025.

Builds on

33 of 47 references.

Closest ancestors · 6

Same direction, newest first

  1. 2024ReplanVLM: Replanning Robotic Tasks with Visual Language Models31 Jul
  2. 2024GameVLM: A Decision-making Framework for Robotic Task Planning Based on Visual Language Models and Zero-sum Games22 May
  3. 2024Yell At Your Robot: Improving On-the-Fly from Language Corrections19 Mar
  4. 2023GPT-4V(ision) for Robotics: Multimodal Task Planning from Human Demonstration20 Nov
  5. 2023Physically Grounded Vision-Language Models for Robotic Manipulation5 Sep
  6. 2023PaLM-E: An Embodied Multimodal Language Model6 Mar

Other references · 27

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  5. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  6. 2023Improved Baselines with Visual Instruction Tuningcited by 4,362
  7. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
  8. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  9. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,674
  10. 2022InstructPix2Pix: Learning to Follow Image Editing Instructionscited by 2,591
  11. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
  12. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
15 more
  1. 2023VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Modelscited by 733
  2. 2023Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Modelscited by 264
  3. 2023PaLI-X: On Scaling up a Multilingual Vision and Language Modelcited by 254
  4. 2018HG-DAgger: Interactive Imitation Learning with Human Expertscited by 245
  5. 2023Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategiescited by 236
  6. 2024How Far is Video Generation from World Model: A Physical Law Perspectivecited by 204
  7. 2025Do generative video models understand physical principles?cited by 151
  8. 2019A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithmscited by 136
  9. 2024FMB: a Functional Manipulation Benchmark for Generalizable Robotic Learningcited by 79
  10. 2024Vision-Language Models Can Self-Improve Reasoning via Reflectioncited by 66
  11. 2024The Benefits of a Concise Chain of Thought on Problem-Solving in Large Language Modelscited by 62
  12. 2024ExACT: Teaching AI Agents to Explore with Reflective-MCTS and Exploratory Learningcited by 61
  13. 2023Multi-Stage Cable Routing through Hierarchical Imitation Learningcited by 59
  14. 2020Learning compositional models of robot skills for task and motion planningcited by 47
  15. 2020Deep Visual Reasoning: Learning to Predict Action Sequences for Task and Motion Planning from an Initial Scene Imagecited by 45

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 20; 2026: 20.

40 papers.

Sort:

By area:

  1. 2025A Survey on Vision-Language-Action Models: An Action Tokenization Perspectivecited by 96
  2. 2025Look-Back: Implicit Visual Re-focusing in MLLM Reasoningcited by 54
  3. 2025Large Language Models for Planning: A Comprehensive and Systematic Surveycited by 38
  4. 2025Agentic Robot: A Brain-Inspired Framework for Vision-Language-Action Models in Embodied Agentscited by 30
  5. 2026Visual Generation Unlocks Human-Like Reasoning through Multimodal World Modelscited by 18
  6. 2025AsyncVLA: Asynchronous Flow Matching for Vision-Language-Action Modelscited by 16
  7. 2026Current Agents Fail to Leverage World Model as Tool for Foresightcited by 15
  8. 2026CycleVLA: Proactive Self-Correcting Vision-Language-Action Models via Subtask Backtracking and Minimum Bayes Risk Decodingcited by 13
  9. 2025Improving Pre-Trained Vision-Language-Action Policies with Model-Based Searchcited by 12
  10. 2026LiLo-VLA: Compositional Long-Horizon Manipulation via Linked Object-Centric Policiescited by 12
  11. 2026Learning from Trials and Errors: Reflective Test-Time Planning for Embodied LLMscited by 10
  12. 2025REMAC: Self-Reflective and Self-Evolving Multi-Agent Collaboration for Long-Horizon Robot Manipulationcited by 8
28 more
  1. 2025Maestro: Orchestrating Robotics Modules with Vision-Language Models for Zero-Shot Generalist Robotscited by 8
  2. 2025PhyBlock: A Progressive Benchmark for Physical Understanding and Planning via 3D Block Assemblycited by 7
  3. 2026Act-Observe-Rewrite: Multimodal Coding Agents as In-Context Policy Learners for Robot Manipulationcited by 7
  4. 2026ETA: A New Agentic Paradigm for Embodied Taskscited by 6
  5. 2025Simulation to Rules: A Dual-VLM Framework for Formal Visual Planningcited by 4
  6. 2026Seeing Farther and Smarter: Value-Guided Multi-Path Reflection for VLM Policy Optimizationcited by 4
  7. 2026Recursive Belief Vision Language Action Modelscited by 3
  8. 2025TRACE: A Self-Improving Framework for Robot Behavior Forecasting with Vision-Language Modelscited by 2
  9. 2026From Instruction to Event: Sound-Triggered Mobile Manipulationcited by 2
  10. 2026PhysMem: Scaling Test-Time Memory for Embodied Physical Reasoningcited by 2
  11. 2026From Imagined Futures to Executable Actions: Mixture of Latent Actions for Robot Manipulationcited by 2
  12. 2025Risk-Sensitive Agent Compositionscited by 1
  13. 2025StepWrite: Adaptive Planning for Speech-Driven Text Generationcited by 1
  14. 2025Long-Horizon Visual Imitation Learning via Plan and Code Reflectioncited by 1
  15. 2025RoboGPT-R1: Enhancing Robot Task Planning with Reinforcement Learningcited by 1
  16. 2025Embodied Tree of Thoughts: Deliberate Manipulation Planning with Embodied World Modelcited by 1
  17. 2026ROSClaw: A Hierarchical Semantic-Physical Framework for Heterogeneous Multi-Agent Collaborationcited by 1
  18. 2026Task-Aware Scanning Parameter Configuration for Robotic Inspection Using Vision Language Embeddings and Hyperdimensional Computingcited by 1
  19. 2026DreamAvoid: Critical-Phase Test-Time Dreaming to Avoid Failures in VLA Policiescited by 1
  20. 2026PhysReflect-VLA: Physical Feasibility and Self-Reflective Regulation for Reliable Vision-Language-Action Policiescited by 1
  21. 2026When Replanning Becomes the Bottleneck: Budgeted Replanning for Embodied Agentscited by 1
  22. 2025Beyond Static Perception: Integrating Temporal Context into VLMs for Cloth Foldingcited by 0
  23. 2025SAGE: Scene Graph-Aware Guidance and Execution for Long-Horizon Manipulation Taskscited by 0
  24. 2025CroSTAta: Cross-State Transition Attention Transformer for Robotic Manipulationcited by 0
  25. 2025Using Language Models as Closed-Loop High-Level Planners for Robotics Applications: A Brief Overview and Benchmarkscited by 0
  26. 2026Dive into the Scene: Breaking the Perceptual Bottleneck in Vision-Language Decision Making via Focus Plan Generationcited by 0
  27. 2026APIVOT: Adaptive Planning with Interleaved Vision-Language Thoughtscited by 0
  28. 2026Triplet2Track: A Hierarchical System with Object-Centric Representations for Reliable Long-Horizon Manipulationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025RoboCerebra: A Large-scale Benchmark for Long-horizon Robotic Manipulation Evaluation
  2. 2026Spatially Grounded Long-Horizon Task Planning in the Wild
  3. 2026Goal2Skill: Long-Horizon Manipulation with Adaptive Planning and Reflection
  4. 2025Large VLM-based Vision-Language-Action Models for Robotic Manipulation: A Survey
  5. 2026Thinking in Text and Images: Interleaved Vision--Language Reasoning Traces for Long-Horizon Robot Manipulation
  6. 2026NovaPlan: Zero-Shot Long-Horizon Manipulation via Closed-Loop Video Language Planning
  7. 2025Reinforced Embodied Planning with Verifiable Reward for Real-World Robotic Manipulation
  8. 2025SIMPACT: Simulation-Enabled Action Planning using Vision-Language Models