Home

Awesome

#! https://zhuanlan.zhihu.com/p/143392167

如需转发，烦请邮件告知 junjialiu@sjtu.edu.cn

Reinforcement-Learning-in-Robotics Content 专栏目录

This is a private learning repository about Reinforcement learning techniques, Reasoning, and Representation learning used in Robotics, founded for Real intelligence.

Reinforcement Learning Foundation

  1. 神经网络基础：反向传播推导与卷积公式 [Zhihu]
  2. 强化学习基础 Ⅰ：马尔可夫与值函数 [Zhihu]
  3. 强化学习基础 Ⅱ：动态规划，蒙特卡洛，时序差分 [Zhihu]
  4. 强化学习基础 Ⅲ：on-policy, off-policy & Model-based, Model-free & Rollout [Zhihu]
  5. 强化学习基础 Ⅳ：State-of-the-art 强化学习经典算法汇总 [Zhihu]
  6. 强化学习基础 Ⅴ：Q learning 原理与实战 [Zhihu]
  7. 强化学习基础 Ⅵ：DQN 原理与实战 [Zhihu]
  8. 强化学习基础 Ⅶ：Double DQN & Dueling DQN 原理与实战 [Zhihu]
  9. 强化学习基础 Ⅷ：Vanilla Policy Gradient 策略梯度原理与实现 [Zhihu]
  10. 强化学习基础 Ⅸ：一文读懂 TRPO 原理与实现 [Zhihu]
  11. 强化学习基础 Ⅹ：一文读懂两种 PPO 原理与实现 [Zhihu]
  12. 强化学习基础 Ⅺ： Actor-Critic & A2C 原理与实现 [Zhihu]
  13. 强化学习基础 Ⅻ：DDPG 原理与实现 [Zhihu]
  14. 强化学习基础 XIII：Twin Delayed DDPG TD3原理与实现 [Zhihu]

Model-based RL

  1. Model-Based RL Ⅰ：Dyna, MVE & STEVE [Zhihu]
  2. Model-Based RL Ⅱ：MBPO原理解读 [Zhihu]
  3. Model-Based RL Ⅲ：从源码读懂PILCO [Zhihu]

Probabilistic in Robotics

  1. PR 序：机器人学的概率方法学习路径 [Zhihu]
  2. PR Ⅰ：最大似然估计MLE与最大后验概率估计MAP [Zhihu]
  3. PR Ⅱ：贝叶斯估计/推断及其与MAP的区别 [Zhihu]
  4. PR Ⅲ：从高斯分布到高斯过程、高斯过程回归、贝叶斯优化 [Zhihu]
  5. PR Ⅳ：贝叶斯神经网络 Bayesian Neural Network [Zhihu]
  6. PR Ⅴ：熵、KL散度、交叉熵、JS散度及python实现 [Zhihu]
  7. PR Ⅵ：多元连续高斯分布的KL散度及python实现 [Zhihu]
  8. PR Sampling Ⅰ：蒙特卡洛采样、重要性采样及python实现 [Zhihu]
  9. PR Sampling Ⅱ：马尔可夫链蒙特卡洛 MCMC及python实现 [Zhihu]
  10. PR Sampling Ⅲ：M-H and Gibbs 采样 [Zhihu]
  11. PR Structured Ⅰ：Graph Neural Network: An Introduction Ⅰ [Zhihu]
  12. PR Structured Ⅱ：Structured Probabilistic Model 结构化概率模型 [Zhihu]
  13. PR Structured Ⅲ：马尔可夫、隐马尔可夫 HMM 、条件随机场 CRF 全解析及其python实现 [Zhihu]
  14. PR Structured Ⅳ：General / Graph Conditional Random Field (CRF) 及其 python 实现 [Zhihu]
  15. PR Structured Ⅴ：GraphRNN——将图生成问题转化为序列生成 [Zhihu]
  16. PR Reasoning 序：Reasoning Robotics 推理机器人学习路线与资源汇总 [Zhihu]
  17. PR Reasoning Ⅰ：Bandit问题与 UCB / UCT / AlphaGo [Zhihu]
  18. PR Reasoning Ⅱ：Relational Inductive bias 关系归纳偏置及其在深度学习中的应用 [Zhihu]
  19. PR Reasoning Ⅲ：基于图表征的关系推理框架 —— Graph Network [Zhihu]
  20. PR Reasoning Ⅳ：数理逻辑（命题逻辑、谓词逻辑）知识整理 [Zhihu]
  21. PR Memory Ⅰ：Memory systems 2018 – towards a new paradigm 【重磅综述】记忆系统——神经科学的启示 [Zhihu]
  22. PR Perspective Ⅰ：Embodied AI 的新浪潮 —— new generation of AI [Zhihu]
  23. PR Perspective Ⅱ：2021/08/03 近期 Robot Learning 领域大事件及思考 [Zhihu]
  24. PR Efficient Ⅰ：机器人中的数据高效强化学习 [Zhihu]
  25. PR Efficient Ⅱ：Bayesian Transfer RL with prior knowledge [Zhihu]
  26. PR Efficient Ⅲ：像训练狗狗一样高效地训练机器人 [Zhihu]
  27. PR Efficient Ⅳ：五分钟内让四足机器人自主学会行走 [Zhihu]
  28. PR Efficient Ⅴ：自预测表征，让RL agent高效地理解世界 [Zhihu]

Meta-Learning

  1. Meta-Learning: An Introduction Ⅰ [Zhihu]
  2. Meta-Learning: An Introduction Ⅱ [Zhihu]
  3. Meta-Learning: An Introduction Ⅲ [Zhihu]

Imitation Learning

  1. Imitation Learning Ⅰ：模仿学习 (Imitation Learning) 入门指南 [Zhihu]
  2. Imitation Learning Ⅱ：DAgger透彻理论分析 [Zhihu]
  3. Imitation Learning Ⅲ：EnsembleDAgger 一种贝叶斯DAgger [Zhihu]

RL from Demonstrations

  1. RLfD Ⅰ：Deep Q-learning from Demonstrations 解读 [Zhihu]
  2. RLfD Ⅱ：Reinforcement Learning from Imperfect Demonstrations under Soft Expert Guidance [Zhihu]

Multi-agent Reinforcement Learning

  1. MARL Ⅰ：A Selective Overview of Theories and Algorithms 【重磅综述】 多智能体强化学习算法理论研究 [Zhihu]

Paper Reading

Active Visual Navigation

  1. Reading：利用物体关系的目标驱动视觉导航 [Zhihu]
  2. Reading：Learning to learn how to learn-Meta自适应视觉导航 [Zhihu]
  3. Reading：Bayesian Relational Memory 在视觉导航中的应用 [Zhihu]
  4. Reading：Attention+3D空间关系图在视觉导航中的应用 [Zhihu]
  5. Reading：机器人导航的半参数化拓扑记忆结构 [Zhihu]
  6. Reading：将Transformer应用到机器人视觉导航中 [Zhihu]

RL robotics in the physical world with micro-data / data-efficiency

  1. 【重磅综述】如何在少量尝试下学习机器人强化学习控制 [Zhihu]

Others

  1. End-to-End Robotic Reinforcement Learning without Reward Engineering: [Medium] [Zhihu]
  2. Overcoming Exploration in RL with Demonstrations: [Medium] [Zhihu]
  3. The Predictron: End-To-End Learning and Planning: [Zhihu]
  4. IROS2019 Paper速读(一) [Zhihu]
  5. IROS2019 Paper速读(二) [Zhihu]
  6. IROS2019 Paper速读(三) [Zhihu]
  7. IROS2019 Paper速读(四) [Zhihu]

Simulator

  1. MuJoCo自定义机器人建模指南
  2. Sim2real in Robotics: An Introduction

Tools

  1. Tools 1：如何用 PyQt5 和 Qt Designer 在 Pycharm 中愉快地开发软件 [Zhihu]
  2. Tools 2：Arxiv 论文提交流程——看这篇就够了 [Zhihu]
  3. Tools 3：Python socket 服务器与客户端双向通信（服务器NAT，文件传输） [Zhihu]
  4. Tools 4：Python三行转并行——真香！[Zhihu]
  5. Tools 5：Python三行转并行后续——多进程全局变量 [Zhihu]

专栏关联Github代码库

Reinforcement-Learning-in-Robotics
Machine-Learning-is-ALL-You-Need

If you're interested in reinforcement learning, we encourage you to check out our latest library of reinforcement learning and imitation learning in (humanoid) robotics.

Release License Documentation Status Build Status

Repository address: https://github.com/Skylark0924/Rofunc