mjlab

MuJoco Simulation on Web Assembly with Neural netwroks

Identify and simulate extended friction models for servo-actuators

Train a Unitree G1 humanoid to perform a double spin kick using mjlab

An open-source unsupervised RL framework for humanoid control with FB/TeCH training, robot-aware motion import, and real-world teleoperation.

One policy. Any motion. Train universal whole-body tracking in mjlab—swap clips, not checkpoints.

AI coding skill kit for IsaacLab migration and mjlab-native development

A collection of tasks built on mjlab

Integrating a custom robot (ANYmal C) with mjlab's velocity task

mjlab reinforcement learning for the BDX-R robot

A growing collection of manipulation tasks built with mjlab.

MuGS: MuJoCo Gaussian Splatting

Dexhand manipulation tasks build upon mjlab

Extension of RSL-RL for using Morphological Symmetries in IsaacLab

Lightweight, GPU-accelerated reinforcement learning for Isaac Lab, mjlab, and Gymnasium.

Predictive Style Matching (PSM), is method in which an offline predictor maps the robot’s lower-body state history and velocity commands to interpretable upper-body joint and gait targets that shape the rewards during training.

Extension of RSL-RL for using Mixture-of-Experts in IsaacLab

The project template for Embodied AI (CS2810 Sp26) at ShanghaiTech

🤖 Teach the Unitree G1 humanoid to perform a double spin kick using mjlab, with training data, pretrained models, and deployment instructions.

RL training of SO-ARM100 in MuJoCo via mjlab

mjlab based drone warfare RL