MJLab (MuJoCo Lab)

MJLab is a GPU-accelerated robotics simulation framework built on MuJoCo (via Warp). It provides vectorized environments running entirely on GPU with fast parallel physics.

Setup

pip install -e ".[mujoco]"
pip install mjlab

How to run

Go1 Velocity (flat terrain)

python run_mjlab.py --config rl_games/configs/mjlab/ppo_go1_velocity.yaml

G1 Humanoid Velocity (flat terrain)

python run_mjlab.py --config rl_games/configs/mjlab/ppo_g1_velocity.yaml

Configs

Environment	Config	Envs	Horizon	Epochs
Go1 Velocity (flat)	configs/mjlab/ppo_go1_velocity.yaml	1024	16	3000
G1 Velocity (flat)	configs/mjlab/ppo_g1_velocity.yaml	1024	32	3000

Results

Go1 Flat Velocity

1024 parallel envs, ~57k FPS on RTX 5090. Converges to reward ~75 within 1000 epochs.

Go1 Rough Velocity

Central value network significantly improves rough terrain performance (~60 vs ~45 reward).

G1 Humanoid Flat Velocity

RSL-style config (v2) with separate actor-critic and entropy 0.001 reaches reward ~65. Baseline config with shared network reaches ~11.