让机器人记忆动态适配当前状态，提升智能体表现与运行效率
MemCompiler: Compile, Don't Inject -- State-Conditioned Memory for Embodied Agents
- 记忆不提前注入，而是根据当前状态动态编译相关记忆
- 在多个环境上比无记忆还高129%，延迟降低60%
- 适合追求高效智能体的开发者和研究者
现有具身智能体的记忆系统通常在任务开始时静态注入记忆，我们称之为事前单体记忆注入（AMMI）。但这种静态设计很快与智能体状态脱节，可能导致轻量级执行器性能低于无记忆基线。为此，我们提出MemCompiler，将记忆使用重构为状态感知的记忆编译。一个学习得到的记忆编译器读取结构化简要状态，动态选择并编译出相关记忆，生成可执行指引。该指引通过文本通道和潜在的软记忆通道传递，后者保留文本无法表达的感知信息。在Alf World、EmbodiedBench和ScienceWorld三个基准上，MemCompiler持续优于无记忆基线（最高提升129%），达到或接近前沿闭源系统水平，并将每步延迟降低60%，证明状态感知的记忆编译能同时提升有效性和效率。
原文摘要 · Abstract (English)
Existing memory systems for embodied agents typically inject retrieved memory as static context at episode start, a paradigm we term Ahead-of-time Monolithic Memory Injection (AMMI). However, this static design quickly becomes misaligned with the agent's evolving state and may degrade lightweight executors below the no-memory baseline. To address this, we propose MemCompiler, which reframes memory utilization as State-Conditioned Memory Compilation. A learned Memory Compiler reads a structured Brief State capturing the agent's current execution state and dynamically selects and compiles only relevant memory into executable guidance. This guidance is delivered through a text channel and a latent Soft-Mem channel that preserves perceptual information not expressible in text. Across Alf World, EmbodiedBench, and ScienceWorld, MemCompiler consistently improves over no-memory across open-source backbones (up to +129%), matches or approaches frontier closed-source systems, and reduces per-step latency by 60%, demonstrating that state-aware memory compilation improves both effectiveness and efficiency.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。