robotmem

概述

robotmem是一个专为AI机器人设计的MCP服务器，用于存储和检索经验数据。它采用混合搜索技术结合BM25全文搜索和向量嵌入，支持通过JSON上下文过滤和空间排序进行结构化经验检索。系统完全基于CPU运行，使用本地SQLite数据库，无需GPU基础设施，特别适用于对空间关系有要求的机器人应用场景。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• learnlearn(insight, context)

  Record physical experiences (parameters / strategies / lessons)

  试试:Record that grip force of 12.5N yields highest grasp success rate试试:Learn from this experiment that pushing with force 15N worked best
• recallrecall(query, context_filter, spatial_sort)

  Retrieve experiences using BM25 + vector hybrid search

  试试:Find experiences with successful push tasks试试:Retrieve memories about grasping objects at position [1.3, 0.7, 0.42]试试:Show me successful experiments with low final distance
• save_perceptionsave_perception(description, perception_type, data)

  Store perception / trajectory / force data

  试试:Save the grasp trajectory data from this successful attempt试试:Store this tactile perception of a soft object
• forgetforget(memory_id)

  Delete incorrect memories

  试试:Remove this failed experiment from memory试试:Forget the experiment where the robot dropped the object
• updateupdate(memory_id, new_content)

  Correct memory content

  试试:Update the memory to reflect the correct grip force试试:Correct this memory with the successful parameters
• start_sessionstart_session(context)

  Begin an episode

  试试:Start a new session for the push task试试:Begin episode with robot arm-01 and task: grasp
• end_sessionend_session(session_id)

  End an episode with auto-consolidation

  试试:End the current session and consolidate memories试试:Finish the experiment and trigger proactive recall

可对比工具

安装

pip install robotmem

{
  "mcpServers": {
    "robotmem": {
      "command": "python",
      "args": ["-m", "robotmem", "mcp"]
    }
  }
}

FAQ

robotmem与其他记忆系统有什么不同？

robotmem专为物理机器人设计，具有结构化经验存储、空间检索和上下文过滤等独特功能，这些是通用记忆系统所不具备的。

robotmem是否需要GPU运行？

不需要，robotmem完全基于CPU运行，使用FastEmbed ONNX模型进行向量搜索，无需GPU基础设施。

Hacker News 讨论

开发者社区最近的相关讨论。

• Show HN: I gave my robot physical memory – it stopped repeating mistakes▲ 18💬 4
  帖子 by robotmem · 2026-03-09

robotmem 对比