Humanoid Stable Locomotion
Isaac Lab, PPO, quadruped→humanoid, RMA, CPG, AMP, distillation — Make a humanoid robot stand up and walk
11
Chapters
10
Interactive
~128
Hours
Chapters
Introduction to Humanoid Robots
10-14hWhy Study Humanoid Robots, Humanoid vs Quadruped vs Wheeled, Degrees of Freedom and Joints, Simulation Environments
Kinematics and Dynamics
12-16hForward Kinematics, Inverse Kinematics, Newton-Euler Dynamics, Lagrangian Mechanics
PPO Fundamentals and Quadruped Locomotion
12-16hPPO Algorithm Review, From Cartpole to Ant: A Progressive Approach, ANYmal A1 Quadruped Walking, From Quadruped to Biped: Key Differences
PPO Humanoid Locomotion Control
12-16hRL for Locomotion Control, Observation Space Design, Reward Engineering, From Standing to Walking
Curriculum Learning and Domain Randomization
10-14hWhy Curriculum Learning Is Needed, Curriculum Design, Domain Randomization, Multi-Terrain Walking
RMA Rapid Motor Adaptation
12-16hThe Privileged Information Problem, Asymmetric Actor-Critic, RMA: Rapid Motor Adaptation, Practical Effects of RMA
CPG-RL Gait Generation
10-14hCentral Pattern Generators in Biology, Mathematical Models of CPG, CPG + RL: Core Architecture, CPG Variants
AMP Motion Priors
12-16hWhy Motion Priors Are Needed, Core Architecture of AMP, AMP Training Process, MoCap Data Processing
Knowledge Distillation and Perception
10-14hKnowledge Distillation Review, Perception Systems, Perception Latency Handling, End-to-End vs Modular Perception
Algorithm Evaluation Methodology
8-10hWhy Evaluation Methodology Matters, Random Seeds and Repeated Experiments, Statistical Metrics, Statistical Tests
Capstone Project — PhaseAdapt
20-30hProject Overview, PhaseAdapt Architecture, Approach Selection and Comparison, Training Pipeline
My Notes & Comments
Sign in to view your highlights, bookmarks, and comments across all chapters.