Projects
Give your knowledge a body.
From a quadruped’s first step to a robot arm’s first grasp. Pick a robot and turn ideas into actions you can see.
Quadrupeds
From the first step to walking freely.
Start with kinematics and joint control, build a gait in simulation, then explore reinforcement learning and language-guided navigation.
Bipeds & wheel-legged robots
Find balance. Explore what comes next.
Teach MicroDuck a movement or balance a wheel-legged robot. Build up training step by step and evaluate the resulting policy.

MicroDuck: teach a robot a movement
From interactive reward design to parallel PPO in MuJoCo Warp.
Open tutorialFlamingo wheel-legged reinforcement learning
6 chapters on Isaac Lab, PPO / CaT, and cross-simulator validation in MuJoCo.
MicroDuck wall-jumping curriculum
12 steps connecting running, jumping, rolling, and standing up.
MicroDuck experiments and creative motions
Explore fixed-condition evaluations and experiments with different motions.
Robot arms
Put your ideas in a robot’s hands.
Connect simulation, demonstrations, and real robot arms through continuous control and imitation learning.

ACT bimanual training
Learn action chunking and closed-loop evaluation with ALOHA demonstrations.
Open tutorialGetting started with MuJoCo
Load an MJCF model and run your first Python control loop.
DDPG continuous-control experiments
π₀.₅ + RECAP in LIBERO
A reproduction path through value models, advantage labels, and policy fine-tuning.
LeRobot course notes
Units 0–2: robot learning, datasets, and tooling.
SO-101 + LeRobot on real hardware
From hardware connections and basic checks to your first motion replay.
AMD projects
Explore another computing platform.
Find cloud environments, ROCm training, and platform-specific robotics guides in one place.
Where ideas meet practice.
When a concept needs explaining, find the next piece in the knowledge map.
Explore foundationsThese overview pages are translated; some tutorials remain in Chinese and display a translation notice.
