mlsanityData to 7 Oct 2026

Reasoning and chain-of-thought in VLA

137 papers1 this weekMedium confidence in this grouping: mostly one topic, with blurry edgesMedium confidence in this grouping: mostly one topic, with blurry edges · About

Keywords: reasoning, memory, vla, manipulation, long-horizon

Key facts. Reasoning and chain-of-thought in VLA: 137 papers on arXiv since 2024 (data through 7 Oct 2026). 56 of its 137 mapped papers (41%) cite “CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models” (2025, arXiv 2503.22020). 54 of its 137 mapped papers (39%) cite “Robotic Control via Embodied Chain-of-Thought Reasoning” (2024, arXiv 2407.08693). 15 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

15 new papers, 17 Sep – 7 Oct; 1 this week.

  1. This week · 1
  2. TempoBridge: Language-Guided Tempo Control for Vision-Language-Action Policies also fits: Generalist VLA models and benchmarks
  3. Earlier · 14
  4. EvoMem-VLA: State-Evolution Memory for Long-Horizon Robot Manipulation also fits: Efficient VLA inference and adaptation
  5. PerturBot: Breaking Shortcut Priors in Vision-Language-Action Models with Perturbative Training also fits: Failure recovery and robustness
  6. Grounded in Time: A Multi-Source Dataset and Benchmark for Temporal Grounding in Robotic Manipulation also fits: Failure recovery and robustness
  7. Divide-and-Remember: Recursive Action-Relevant Memory for Long-Horizon VLA Policies also fits: Efficient VLA inference and adaptation
  8. When Reasoning Helps Action: Monitoring and Steering Chain-of-Thought in Vision-Language-Action Policies also fits: Failure recovery and robustness
  9. Inline Memory Meets Reusable Skills: Memory-centric Framework for Vision-Language-Action Model also fits: Reinforcement learning and post-training
  10. D2-VLA: Dual-Memory Dual-Frequency Vision-Language-Action Model For Long Dynamic Manipulation also fits: Efficient VLA inference and adaptation
  11. Remember What You Did: Action-History Memory with Dual-Expert Denoising for Long-Horizon Vision-Language-Action Policies also fits: Efficient VLA inference and adaptation
  12. Simple Agentic Memory for Generalist Robot Policies also fits: Failure recovery and robustness
All 15
  1. RoboIRGBench: Benchmarking Implicit Referential Grounding in Vision-Language-Action Models also fits: Failure recovery and robustness
  2. Resolving State-Representation Mismatch: State-Space Visual Reasoning for Open-Loop VLA Planning also fits: Efficient VLA inference and adaptation
  3. ActionGround: Training-Free Runtime Refinement of Frozen VLA Policies also fits: Generalist VLA models and benchmarks
  4. MemBodied: Recurrent Associative Memory for Vision-Language-Action Models also fits: Failure recovery and robustness
  5. X-Planner: Event-Structured Task Planning for Embodied Intelligence

Foundations

Most cited by papers here

  1. CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models2025cited by 56 here
  2. Robotic Control via Embodied Chain-of-Thought Reasoning2024cited by 54 here
  3. MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation2025cited by 37 here
  4. ThinkAct: Vision-Language-Action Reasoning via Reinforced Visual Latent Planning2025cited by 28 here
  5. Training Strategies for Efficient Embodied Reasoning2025cited by 16 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 91 here
  2. 2024Octo: An Open-Source Generalist Robot Policycited by 64 here
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 110 here
  4. 2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulationcited by 16 here
  5. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 97 here
  6. 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 42 here
  7. 2025SpatialVLA: Exploring Spatial Representations for Visual-Language-Action Modelcited by 44 here
  8. 2025MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulationcited by 37 here
  9. 2025X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Modelcited by 17 here
15 more it builds on, not in this chain
  1. 2024Robotic Control via Embodied Chain-of-Thought Reasoningcited by 54 here
  2. 2024CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulationcited by 29 here
  3. 2024TraceVLA: Visual Trace Prompting Enhances Spatial-Temporal Awareness for Generalist Robotic Policiescited by 45 here
  4. 2025HAMSTER: Hierarchical Action Models For Open-World Robot Manipulationcited by 16 here
  5. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 57 here
  6. 2025CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Modelscited by 56 here
  7. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 84 here
  8. 2025GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Datacited by 14 here
  9. 2025UniVLA: Learning to Act Anywhere with Task-centric Latent Actionscited by 19 here
  10. 2025Training Strategies for Efficient Embodied Reasoningcited by 16 here
  11. 2025OneTwoVLA: A Unified Vision-Language-Action Model with Adaptive Reasoningcited by 16 here
  12. 2025SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Roboticscited by 23 here
  13. 2025DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledgecited by 16 here
  14. 2025ThinkAct: Vision-Language-Action Reasoning via Reinforced Visual Latent Planningcited by 28 here
  15. 2025HAMLET: Switch your Vision-Language-Action Model into a History-Aware Policycited by 16 here

The 24 most cited of 370.

Browse

30 of 137, ranked 6–35

  1. 6OneTwoVLA: A Unified Vision-Language-Action Model with Adaptive Reasoning202516
  2. 7HAMLET: Switch your Vision-Language-Action Model into a History-Aware Policy202516
  3. 8MEM: Multi-Scale Embodied Memory for Vision Language Action Models202614
  4. 9ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action Model202513
  5. 10MAP-VLA: Memory-Augmented Prompting for Vision-Language-Action Model in Robotic Manipulation202511
  6. 11RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies202610
  7. 12InstructVLA: Vision-Language-Action Instruction Tuning from Understanding to Manipulation202510
  8. 13DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models202510
  9. 14Fast-ThinkAct: Efficient Vision-Language-Action Reasoning via Verbalizable Latent Planning20268
  10. 15ReMem-VLA: Empowering Vision-Language-Action Model with Memory via Dual-Level Recurrent Queries20267
All 30
  1. 16CoA-VLA: Improving Vision-Language-Action Models via Visual-Textual Chain-of-Affordance20247
  2. 17Do What You Say: Steering Vision-Language-Action Models via Runtime Reasoning-Action Alignment Verification20256
  3. 18Rethinking Progression of Memory State in Robotic Manipulation: An Object-Centric Perspective20256
  4. 19ACoT-VLA: Action Chain-of-Thought for Vision-Language-Action Models20265
  5. 20Latent Reasoning VLA: Latent Thinking and Prediction for Vision-Language-Action Models20265
  6. 21LaST0: Latent Spatio-Temporal Chain-of-Thought for Robotic Vision-Language-Action Model20264
  7. 22Non-Markovian Long-Horizon Robot Manipulation via Keyframe Chaining20264
  8. 23DualCoT-VLA: Visual-Linguistic Chain of Thought via Parallel Reasoning for Vision-Language-Action Models20264
  9. 24Long-Horizon Manipulation via Trace-Conditioned VLA Planning20264
  10. 25Fast ECoT: Efficient Embodied Chain-of-Thought via Thoughts Reuse20254
  11. 26FlowVLA: Visual Chain of Thought-based Motion Reasoning for Vision-Language-Action Models20254
  12. 27ManualVLA: A Unified VLA Model for Chain-of-Thought Manual Generation and Robotic Manipulation20254
  13. 28Action-Sketcher: From Reasoning to Action via Visual Sketches for Long-Horizon Robotic Manipulation20263
  14. 29TempoFit: Plug-and-Play Layer-Wise Temporal KV Memory for Long-Horizon Vision-Language-Action Manipulation20263
  15. 30Beyond Short-Horizon: VQ-Memory for Robust Long-Horizon Manipulation in Non-Markovian Simulation Benchmarks20263
  16. 31Goal2Skill: Long-Horizon Manipulation with Adaptive Planning and Reflection20263
  17. 32HELM: Harness-Enhanced Long-horizon Memory for Vision-Language-Action Manipulation20263
  18. 33EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies20263
  19. 34PALM: Progress-Aware Policy Learning via Affordance Reasoning for Long-Horizon Robotic Manipulation20262
  20. 35BagelVLA: Enhancing Long-Horizon Manipulation via Interleaved Vision-Language-Action Generation20262