# thousandbrainsproject

**Repository Path**: wudancome/thousandbrainsproject

## Basic Information

- **Project Name**: thousandbrainsproject
- **Description**: 千脑系统：面向快速、鲁棒学习与推理的感觉运动智能
当前的人工智能系统在诸多任务上取得了令人瞩目的性能，但它们仍欠缺生物智能的核心特质，包括快速、持续的学习能力、基于感觉运动交互形成的表征，以及能实现高效泛化的结构化知识。神经科学理论表明，哺乳动物通过复制一种半独立的感觉运动模块 —— 一种被称为皮质柱的功能单元 —— 进化出了灵活的智能。为弥合生物智能与人工智能之间的差距，研究者提出了千脑系统，以
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-03-11
- **Last Updated**: 2026-03-11

## Categories & Tags

**Categories**: Uncategorized

**Tags**: 感觉运动, 世界模型, 具身智能, 主动感知, 参考系

## README

# 一、总览：概述
- [1. 欢迎阅读千脑计划文档](overview/welcome-to-the-thousand-brains-project-documentation.md)
- [2. 千脑计划愿景](overview/vision-of-the-thousand-brains-project.md)
  - [2.1 长期目标与原则](overview/vision-of-the-thousand-brains-project/long-term-goals-and-principles.md)
  - [2.2 短期目标](overview/vision-of-the-thousand-brains-project/short-term-goals.md)
  - [2.3 打破固有认知](overview/vision-of-the-thousand-brains-project/challenging-preconceptions.md)
  - [2.4 系统能力](overview/vision-of-the-thousand-brains-project/capabilities-of-the-system.md)
- [3. 架构概览](overview/architecture-overview.md)
  - [3.1 传感器模块](overview/architecture-overview/sensor-modules.md)
  - [3.2 学习模块](overview/architecture-overview/learning-modules.md)
  - [3.3 皮层通信协议](overview/architecture-overview/cortical-messaging-protocol.md)
  - [3.4 其他方面](overview/architecture-overview/other-aspects.md)
  - [3.5 整体整合](overview/architecture-overview/bringing-it-together.md)
- [4. 应用标准](overview/application-criteria.md)
- [5. 基准实验](overview/benchmark-experiments.md)
  - [5.1 替代方案实现结果](overview/benchmark-experiments/results-from-alternative-implementations.md)
- [6. 千脑计划常见问题](overview/faq-thousand-brains-project.md)
- [7. 术语表](overview/glossary.md)
- [8. 延伸阅读](overview/further-reading.md)
  - [8.1 社区与媒体报道](overview/further-reading/community-and-media-coverage.md)

# 二、如何使用 Monty
- [1. 快速上手](how-to-use-monty/getting-started.md)
  - [1.1 Windows 环境通过 WSL 快速上手](how-to-use-monty/getting-started/getting-started-on-windows-via-wsl.md)
- [2. 代码库结构](how-to-use-monty/code-base-structure.md)
- [3. 运行基准测试](how-to-use-monty/running-benchmarks.md)
- [4. 日志与分析](how-to-use-monty/logging-and-analysis.md)
- [5. 自定义 Monty](how-to-use-monty/customizing-monty.md)
- [6. 实现动作逻辑](how-to-use-monty/implementing-actions.md)
- [7. 教程](how-to-use-monty/tutorials.md)
  - [7.1 运行你的第一个实验](how-to-use-monty/tutorials/running-your-first-experiment.md)
  - [7.2 模型预训练](how-to-use-monty/tutorials/pretraining-a-model.md)
  - [7.3 基于预训练模型执行推理](how-to-use-monty/tutorials/running-inference-with-a-pretrained-model.md)
  - [7.4 无监督持续学习](how-to-use-monty/tutorials/unsupervised-continual-learning.md)
  - [7.5 多学习模块](how-to-use-monty/tutorials/multiple-learning-modules.md)
  - [7.6 在自定义应用中使用 Monty](how-to-use-monty/tutorials/using-monty-in-a-custom-application.md)
  - [7.7 Monty 在机器人领域的应用](how-to-use-monty/tutorials/using-monty-for-robotics.md)
  - [7.8 从其他代码仓库运行实验](how-to-use-monty/tutorials/running-an-experiment-from-a-different-repository.md)
- [8. 常见问题及解决方法](how-to-use-monty/common-issues-and-how-to-fix-them.md)

# 三、Monty 工作原理
- [1. 实现概述](how-monty-works/implementation-overview.md)
- [2. 实验](how-monty-works/experiment.md)
- [3. 环境与智能体](how-monty-works/environment-agent.md)
- [4. Monty](how-monty-works/monty.md)
  - [4.1 约定规范](how-monty-works/monty/conventions.md)
  - [4.2 参考系变换](how-monty-works/monty/reference-frame-transformations.md)
- [5. 观测、变换与传感器模块](how-monty-works/observations-transforms-sensor-modules.md)
- [6. 学习模块](how-monty-works/learning-module.md)
  - [6.1 对象模型](how-monty-works/learning-module/object-models.md)
  - [6.2 学习模块输出](how-monty-works/learning-module/learning-module-outputs.md)
  - [6.3 基于证据的学习模块](how-monty-works/learning-module/evidence-based-learning-module.md)
- [7. 将学习模块连接为异层结构](how-monty-works/connecting-lms-into-a-heterarchy.md)
- [8. 策略](how-monty-works/policy.md)
- [9. 研究特性](how-monty-works/research-features.md)
  - [9.1 作为相似度度量的证据](how-monty-works/research-features/evidence-as-a-similarity-measure.md)
- [10. Monty 常见问题](how-monty-works/faq-monty.md)
- [11. 开放问题](how-monty-works/open-questions.md)

# 四、贡献指南
- [1. 为何贡献](contributing/why-contribute.md)
- [2. 参与贡献](contributing/contributing.md)
- [3. 拉取请求](contributing/pull-requests.md)
  - [3.1 贡献者许可协议](contributing/pull-requests/contributor-license-agreement.md)
  - [3.2 拉取请求流程](contributing/pull-requests/pull-request-flow.md)
  - [3.3 拉取请求评审](contributing/pull-requests/reviewing-a-pull-request.md)
- [4. 代码贡献方式](contributing/ways-to-contribute-to-code.md)
  - [4.1 认领待处理问题](contributing/ways-to-contribute-to-code/identify-an-issue-to-work-on.md)
- [5. 文档建设](contributing/documentation.md)
  - [5.1 未来工作组件元数据](contributing/documentation/future-work-widget-metadata.md)
- [6. 贡献教程](contributing/contributing-tutorials.md)
- [7. 问题评论](contributing/comment-on-issues.md)
- [8. 意见征求（RFC）](contributing/request-for-comments-rfc.md)
- [9. 风格指南](contributing/style-guide.md)
  - [9.1 代码风格指南](contributing/style-guide/code-style-guide.md)
  - [9.2 类型注解指南](contributing/style-guide/typing-guide.md)
- [10. 维护者指南](contributing/guides-for-maintainers.md)
  - [10.1 问题分类处理](contributing/guides-for-maintainers/triage.md)
- [11. 大模型使用规范](contributing/large-language-model-use.md)

# 五、社区
- [1. 行为准则](community/code-of-conduct.md)
- [2. 治理规范](community/governance.md)
- [3. 许可证与专利承诺说明](community/our-licence-and-patent-pledge-explained.md)
- [4. 基于千脑计划的论文](community/tbp-based-papers.md)
- [5. 项目展示](community/project-showcase.md)

# 六、理论：神经科学理论
- [1. 理论概述](theory/theory-overview.md)
- [2. 组合模型](theory/compositional-models.md)
- [3. 对象行为](theory/object-behaviors.md)

# 七、未来工作
- [1. 项目路线图](future-work/project-roadmap.md)
- [2. 传感器模块优化](future-work/sensor-module-improvements.md)
  - [2.1 提取更优特征](future-work/sensor-module-improvements/extract-better-features.md)
  - [2.2 表面传感器模块](future-work/sensor-module-improvements/surface-sm.md)
  - [2.3 视差深度感知](future-work/sensor-module-improvements/parallax-for-depth-perception.md)
  - [2.4 局部与全局光流检测](future-work/sensor-module-improvements/detect-local-and-global-flow.md)
  - [2.5 变化检测传感器模块](future-work/sensor-module-improvements/change-detecting-sm.md)
- [3. 学习模块优化](future-work/learning-module-improvements.md)
  - [3.1 贡献学习模块](future-work/learning-module-improvements/contributing-learning-modules.md)
  - [3.2 利用非对象观测数据](future-work/learning-module-improvements/use-off-object-observations.md)
  - [3.3 使用模型外运动信息](future-work/learning-module-improvements/use-out-of-model-movements.md)
  - [3.4 实现并测试快速证据衰减作为无监督记忆重置方式](future-work/learning-module-improvements/implement-and-test-rapid-evidence-decay-as-form-of-unsupervised-memory-resetting.md)
  - [3.5 提升有界证据性能](future-work/learning-module-improvements/improve-bounded-evidence-performance.md)
  - [3.6 使用更少关键点的模型](future-work/learning-module-improvements/use-models-with-fewer-points.md)
  - [3.7 测试类似粒子滤波的假设空间重采样](future-work/learning-module-improvements/test-particle-filter-like-resampling-of-hypothesis-space.md)
  - [3.8 动态调整假设空间大小](future-work/learning-module-improvements/dynamic-adjustment-of-hypothesis-space-size.md)
  - [3.9 采样子假设](future-work/learning-module-improvements/sample-offspring-hypotheses.md)
  - [3.10 重新锚定假设](future-work/learning-module-improvements/re-anchor-hypotheses.md)
  - [3.11 降低对首次观测的依赖](future-work/learning-module-improvements/less-dependency-on-first-observation.md)
  - [3.12 处理不完整模型](future-work/learning-module-improvements/deal-with-incomplete-models.md)
  - [3.13 测试网格对象模型用于无监督学习](future-work/learning-module-improvements/test-grid-object-models-for-unsupervised-learning.md)
  - [3.14 处理运动对象](future-work/learning-module-improvements/deal-with-moving-objects.md)
  - [3.15 支持尺度不变性](future-work/learning-module-improvements/support-scale-invariance.md)
  - [3.16 优化对称性处理](future-work/learning-module-improvements/improve-handling-of-symmetry.md)
  - [3.17 使用更优先验进行假设初始化](future-work/learning-module-improvements/use-better-priors-for-hypothesis-initialization.md)
  - [3.18 学习对象热点区域](future-work/learning-module-improvements/learn-hot-spots-on-objects.md)
  - [3.19 在模型中加入状态](future-work/learning-module-improvements/include-state-in-models.md)
  - [3.20 在假设中加入状态](future-work/learning-module-improvements/include-state-in-hypotheses.md)
  - [3.21 事件检测重置计时器](future-work/learning-module-improvements/event-detection-to-reset-timer.md)
  - [3.22 速度检测调整计时器](future-work/learning-module-improvements/speed-detection-to-adjust-timer.md)
- [4. 运动系统优化](future-work/motor-system-improvements.md)
  - [4.1 实现学习与推理策略切换](future-work/motor-system-improvements/implement-switching-between-learning-and-inference-focused-policies.md)
  - [4.2 表面智能体自底向上探索策略](future-work/motor-system-improvements/bottom-up-exploration-policy-for-surface-agent.md)
  - [4.3 基于模型的探索策略](future-work/motor-system-improvements/model-based-exploration-policy.md)
  - [4.4 实现由无模型与基于模型信号驱动的高效扫视](future-work/motor-system-improvements/implement-efficient-saccades-driven-by-model-free-and-model-based-signals.md)
  - [4.5 通过强化学习学习策略](future-work/motor-system-improvements/learn-policy-using-rl.md)
  - [4.6 目标分解为子目标理论](future-work/motor-system-improvements/theory-for-decomposing-goals-into-subgoals.md)
  - [4.7 在运动系统中解析目标状态并切换策略](future-work/motor-system-improvements/interpret-goal-states-in-motor-system-switch-policies.md)
  - [4.8 识别对象后再切换新对象的策略](future-work/motor-system-improvements/policy-to-recognize-an-object-before-moving-on-to-a-new-object.md)
  - [4.9 快速切换至新对象的策略](future-work/motor-system-improvements/policy-to-quickly-move-to-a-new-object.md)
- [5. 投票机制优化](future-work/voting-improvements.md)
  - [5.1 清理并简化投票逻辑](future-work/voting-improvements/clean-up-and-simplify-voting.md)
  - [5.2 梳理注意力应用方式](future-work/voting-improvements/outline-how-to-apply-attention.md)
  - [5.3 将投票泛化至关联连接](future-work/voting-improvements/generalize-voting-to-associative-connections.md)
  - [5.4 能否将皮层通信协议改为使用位移而非位置](future-work/voting-improvements/can-we-change-the-cmp-to-use-displacements-instead-of-locations.md)
  - [5.5 对状态进行投票](future-work/voting-improvements/vote-on-state.md)
- [6. 皮层通信协议层级优化](future-work/cmp-hierarchy-improvements.md)
  - [6.1 确定异层结构中的性能度量与监督方式](future-work/cmp-hierarchy-improvements/figure-out-performance-measure-and-supervision-in-heterarchy.md)
  - [6.2 添加自上而下连接](future-work/cmp-hierarchy-improvements/add-top-down-connections.md)
  - [6.3 在异层结构测试台中运行并分析更多学习模块的实验](future-work/cmp-hierarchy-improvements/run-analyze-experiments-with-more-lms-in-heterarchy-testbed.md)
  - [6.4 在 Omniglot 数据集上评估](future-work/cmp-hierarchy-improvements/evaluate-on-omniglot-dataset.md)
  - [6.5 测试不同层级的学习速度](future-work/cmp-hierarchy-improvements/test-learning-at-different-speeds-depending-on-level-in-hierarchy.md)
  - [6.6 将相似度编码与对象 ID 发送至下一层级测试](future-work/cmp-hierarchy-improvements/send-similarity-encoding-object-id-to-next-level-test.md)
  - [6.7 全局间隔计时器](future-work/cmp-hierarchy-improvements/global-interval-timer.md)
  - [6.8 在皮层通信协议中加入状态](future-work/cmp-hierarchy-improvements/include-state-in-CMP.md)
- [7. 环境优化](future-work/environment-improvements.md)
  - [7.1 对象行为测试平台](future-work/environment-improvements/object-behavior-test-bed.md)
  - [7.2 构建组合对象测试数据集](future-work/environment-improvements/make-dataset-to-test-compositional-objects.md)
  - [7.3 搭建支持对象操作的环境](future-work/environment-improvements/set-up-environment-that-allows-for-object-manipulation.md)
  - [7.4 搭建对象操作基准任务与评估指标](future-work/environment-improvements/set-up-object-manipulation-benchmark-tasks-and-evaluation-measures.md)
  - [7.5 创建数据集与指标评估类别与泛化能力](future-work/environment-improvements/create-dataset-and-metrics-to-evaluate-categories-and-generalization.md)
  - [7.6 创建数据集测试新特征与形态组合](future-work/environment-improvements/create-dataset-and-metrics-to-test-new-feature-morphology-pairs.md)
  - [7.7 简化仿真器协议](future-work/environment-improvements/simplify-simulator-protocol.md)
- [8. 框架优化](future-work/framework-improvements.md)
  - [8.1 添加多独立运动智能体基础架构](future-work/framework-improvements/add-infrastructure-for-multiple-agents-that-move-independently.md)
  - [8.2 自动化基准实验分析](future-work/framework-improvements/automate-benchmark-experiments-analysis.md)
  - [8.3 为无监督学习添加更多 wandb 日志](future-work/framework-improvements/add-more-wandb-logging-for-unsupervised-learning.md)
  - [8.4 为 Monty 添加 GPU 支持](future-work/framework-improvements/add-gpu-support-for-monty.md)
  - [8.5 在学习模块内部使用状态类](future-work/framework-improvements/use-state-class-inside-of-lms.md)
  - [8.6 简化配置使用](future-work/framework-improvements/make-configs-easier-to-use.md)
  - [8.7 寻找比 KD 树搜索更快的替代方案](future-work/framework-improvements/find-faster-alternative-to-kdtree-search.md)
  - [8.8 将数据加载器与数据集整合为环境接口](future-work/framework-improvements/combine-dataloader-and-dataset-into-environment-interface.md)
- [9. 开源社区沟通优化](future-work/oss-communication-improvements.md)
  - [9.1 在 README 中自动展示贡献者](future-work/oss-communication-improvements/auto-display-contributors-in-readme.md)
  - [9.2 更优的文档平台](future-work/oss-communication-improvements/better-documentation-platform.md)
  - [9.3 更优的官网平台](future-work/oss-communication-improvements/better-website-platform.md)
  - [9.4 制作教程](future-work/oss-communication-improvements/build-a-tutorial.md)
  - [9.5 梳理论坛问题主题并完善文档](future-work/oss-communication-improvements/find-forum-questions-themes-improve-docs.md)
  - [9.6 未来工作组件](future-work/oss-communication-improvements/future-work-widget.md)
  - [9.7 在文档中插入术语表词条](future-work/oss-communication-improvements/insert-glossary-terms-in-docs.md)
  - [9.8 制作更多关于 Monty 项目的精简视频](future-work/oss-communication-improvements/make-more-condensed-videos-about-the-project-monty.md)
  - [9.9 组织线下见面会](future-work/oss-communication-improvements/organize-local-meetup.md)
  - [9.10 启动播客系列节目](future-work/oss-communication-improvements/organize-start-podcast-series.md)
  - [9.11 稳健的文档预览清理](future-work/oss-communication-improvements/robust-docs-preview-cleanup.md)