sim2sim

Human mocap retargeting and whole-body motion tracking for humanoids (Unitree G1, IGRIS-C) in Isaac Lab and MuJoCo. Simulation only.

Unitree G1 locomotion: Isaac Lab RL -> ONNX manifest -> MuJoCo sim-to-sim -> 1 kHz C++ control

GEAR-SONIC adaptation for the 23-DoF Galaxea Kengo; restricted robot assets are pinned through a private submodule.

Modify Tien Nghich Mobile game values and gameplay parameters using this Android script utility.

sim2real of closed-loop quadrupedal robot

Deploy the RL policy in mujoco

Policy-driven robotics simulation experiments with measurable task execution, physics validation, and simulator-to-simulator testing.

The code & files for the article "Metrics for Assessing Generalization of Deep Reinforcement Learning in Parameterized Environments".

Unitree Go2 Tier-1 RoboPay profile — 9 priced embodied skills (wave, sit, stand, stop, bow, nod, turn_to_face, hold, navigate_obstacle) on MuJoCo + PyBullet sim-to-sim (measured ≤0.02 cm) + Webots R2025a harness, x402 payment gate, durable replay, and live Base Sepolia EIP-3009 settlement proof.

Go1 side flip: Isaac Gym training + MuJoCo sim2sim

四足机器人 Go2 全栈控制：Convex MPC · RL（BC/PPO/域随机化）· 混合仲裁，同一套 72 试验推力协议下恢复率 94% —— 全程 CPU 训练，附与代码逐项一致的数学设计文档

DM-Dog-RL is a reinforcement-learning locomotion project for a Damiao-motor quadruped robot. It contains both the training stack and the real-robot ROS deployment workspace.

Navegación autónoma basada en Redes Neuronales

Repository created using https://github.com/google-deepmind/mujoco_playground as base in order to add the Go2 walking and handstand policies. It contains the code to train and simulate the policy.

Velocity tracking and sim-to-sim transfer for Unitree H1 humanoid locomotion using reinforcement learning.

Lightweight repo for getting started with learning to control the Unitree G1 humanoid robot with sim2sim (mujoco & isaac lab) and sim2real policy validation.

Deep Reinforcement Learning and Imitation Learning with sim2sim exploration and behavior cloning

DDC (Deployable Dynamic-CoM): a unified policy and method-agnostic sim2sim benchmark for humanoid single-leg stance; deploys on a real Unitree G1. arXiv:2608.00500

Automatic Domain Randomization (ADR) proposed in "Solving Rubik's Cube with a Robot Hand"

A ROS2 controller for a UR arm leveraging IsaacLab's trained policy for sim-to-real reaching tasks.