mlsanityData to 14 Oct 2026

Recursive Video In-Context Learning for Agentic Robot

arXiv 2610.06843 ↗First posted 5 Oct 2026cs.RO cs.AI cs.CLcited by 0

On the map: Vision-language-action models → Reinforcement learning and post-training also fits: Reasoning and chain-of-thought in VLA

also in 1 more
Abstract (arXiv)

LLM agents that orchestrate frozen vision-language-action (VLA) policies improve across episodes through text memory, which records what the agent did but not how the task is done. A demonstration video shows it, but fits poorly into an agent's context. The full video slows every turn, fixed keyframes lose the contact detail that decides whether a grasp holds, and what the agent needs shifts from the task's structure while planning to the frames around each contact. We introduce Recursive Video In-Context Learning (RV-ICL), a training-free method that turns a demonstration into a hierarchy the agent navigates rather than a prompt it receives. The hierarchy is built from the sub-events of the demonstration, such as grasps and releases. Its levels grow finer, from keyframes of the whole task to phases, moments and short clips, and are exposed through read-only tools. The agent reads the coarse levels before planning. During execution it re-enters the hierarchy whenever a step needs more detail and loads only the clip of its current sub-goal. One demonstration per task is enough. Built on RPent, RV-ICL raises success from 92.6% to 96.5% on LIBERO-PRO and from 86.7% to 95.8% on LIBERO-Plus.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reinforcement learning and post-training, one of 13 directions in Vision-language-action models. Builds on 22 of its 46 references that are in the map. First posted on arXiv 5 Oct 2026.

Builds on

22 of 46 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026AtomVLA: Scalable Post-Training for Robotic Manipulation via Predictive Latent World Models9 Mar

Other references · 21

  1. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  2. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,191
  3. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,761
  4. 2023MemGPT: Towards LLMs as Operating Systemscited by 1,389
  5. 2025LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Modelscited by 303
  6. 2021Recursively Summarizing Books with Human Feedbackcited by 290
  7. 2025MolmoAct: Action Reasoning Models that can Reason in Spacecited by 197
  8. 2025LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorizationcited by 179
  9. 2025NORA: A Small Open-Sourced Generalist Vision Language Action Model for Embodied Taskscited by 127
  10. 2025Recursive Language Modelscited by 104
  11. 2026CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulationcited by 88
  12. 2024Keypoint Action Tokens Enable In-Context Imitation Learning in Roboticscited by 74
9 more
  1. 2024In-Context Imitation Learning via Next-Token Predictioncited by 70
  2. 2025RLinf: Flexible and Efficient Large-scale Reinforcement Learning via Macro-to-Micro Flow Transformationcited by 57
  3. 2024Vid2Robot: End-to-end Video-conditioned Policy Learning with Cross-Attention Transformerscited by 52
  4. 2026Harness VLA: Steering Frozen VLAs into Reliable Manipulation Primitives via Memory-Guided Agentscited by 47
  5. 2026Playful Agentic Robot Learningcited by 22
  6. 2026An Unexpected Robot Policy: Early Evaluations of GPT-6 Astra on RoboDojo and Beyondcited by 11
  7. 2026StellaVLA: In-Context Structured Demonstration for Generalizable Vision-Language-Action Modelscited by 4
  8. 2026Robo-Harness K1: Harnessing Robot-Use Agents via Perception Augmentationcited by 3
  9. 2026RAPID: Robot Agentic Programming from Demonstrationscited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026In-Context Robot Learning with VLM Agents
  2. 2025ContextVLA: Vision-Language-Action Model with Amortized Multi-Frame Context
  3. 2026V-ICAL Bench: Evaluating Video In-Context Learning for Multimodal Agents in Interactive Environments
  4. 2025RICL: Adding In-Context Adaptability to Pre-Trained Vision-Language-Action Models
  5. 2025See Once, Then Act: Vision-Language-Action Model with Task Learning from One-Shot Video Demonstrations
  6. 2026ReMem-VLA: Empowering Vision-Language-Action Model with Memory via Dual-Level Recurrent Queries
  7. 2025Leveraging In-Context Learning for Language Model Agents
  8. 2024VLM Agents Generate Their Own Memories: Distilling Experience into Embodied Programs of Thought