Research map / Vision-language and multimodal / Robot action and navigation
Remembering What Matters: From Markovian to Subtask-Causal Memory in VLA Policies
Peishuo Wang, Fengshuo Bai, Yufeng Li, Tawei Chou, Yinda Xu, Ying Wen, Yaodong Yang, Chen Gao, Zhenzhe Zheng, Fan Wu
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes subtask-causal memory for vision-language-action policies so that long-horizon manipulation can recall earlier subtask state changes beyond recent-frame windows.
Problem
Long-horizon robotic manipulation is non-Markovian: correct actions may depend on earlier state changes no longer visible in the current observation, and VLA policies degrade when they must remember which earlier subtasks changed the world.
Approach
Moves from Markovian and recent-frame-window memory to memory organized around causally relevant subtasks, retaining what earlier subtasks changed rather than raw recent history (from the title and abstract; details not available).
Experimental setup
Baselines: recent-frame window memory
Closest papers by meaning
- VLA-Pro: Cross-Task Procedural Memory Transfer for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation (ICML 2025 · similarity 0.90)
- ECHO: Continuous Hierarchical Memory for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (NeurIPS 2025 · similarity 0.89)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- STRAP: Robot Sub-Trajectory Retrieval for Augmented Policy Learning (ICLR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
