研究库 论文知识库

内容库 / 论文

arXiv:2609.11561 · 工程化

Memory as Plans: World-Action Modeling with Memory-Grounded Planning

以记忆为规划：基于记忆锚定的世界-动作建模规划

Memory as Plans: World-Action Modeling with Memory-Grounded Planning

类型：arxiv
标识：2609.11561
链接：https://arxiv.org/abs/2609.11561
主分类：engineering
形态：method
TLDR：Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their combinations, and may therefore lose fine-grained visual evidence or face a trade-off between history coverage and execution efficiency. We introduce MaP-WAM, a Memory-as-Plans framework that decomposes memory-dependent world-action modeling into memory-grounded planning and plan-conditioned execution
待LLM分类：是
标题中文：以记忆为规划：基于记忆锚定的世界-动作建模规划
TLDR中文：主流机器人策略通常采用马尔可夫式建模，但许多复杂的现实世界操作任务本质上是非马尔可夫的，需要超越当前观测的长期记忆。现有记忆机制往往依赖语言摘要、不断增长的视觉窗口或二者的组合，因而可能丢失细粒度视觉证据，或在历史覆盖范围与执行效率之间难以权衡。我们提出 MaP-WAM，一个将记忆依赖的世界-动作建模分解为记忆锚定规划与规划条件执行的"以记忆为规划"框架……
来源文件：
/inbox/tom/_candidates/2026-09-11-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

笔记 Substack · α-signal · RAG and Long Context Aren't Enough for Agent Memory. δ-mem Is a Third Option

笔记 flyP 轻量精读 · Beyond Memory: A Transactional Continuity Kernel for Long-Lived AI Agents（arXiv:2608.11632 · cs.MA · 8-12 v1）

笔记 Tom 文献雷达 · AI Agent / RAG / 长上下文 · 2026-07-30（午后版）

笔记 Tom 文献雷达 · AI Agent × RAG × Long Context · 2026-06-24

仓库 kolegadev/Katra-Agentic-Memory

笔记 Tom 文献雷达 · Agent + RAG + Long-Context · 2026-08-13

arXiv

2609.11561

TLDR

主流机器人策略通常采用马尔可夫式建模，但许多复杂的现实世界操作任务本质上是非马尔可夫的，需要超越当前观测的长期记忆。现有记忆机制往往依赖语言摘要、不断增长的视觉窗口或二者的组合，因而可能丢失细粒度视觉证据，或在历史覆盖范围与执行效率之间难以权衡。我们提出 MaP-WAM，一个将记忆依赖的世界-动作建模分解为记忆锚定规划与规划条件执行的"以记忆为规划"框架……Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their combinations, and may therefore lose fine-grained visual evidence or face a trade-off between history coverage and execution efficiency. We introduce MaP-WAM, a Memory-as-Plans framework that decomposes memory-dependent world-action modeling into memory-grounded planning and plan-conditioned execution

arXiv ↗ 中文全文