Language

Configuration¶

Domain randomization is configured inside the selected task owner YAML, usually under env.domain_rand. Use --task and --sim to select backend-specific behavior first, then override fields inside that selected owner.

uv run train --algo ppo --task g1_walk_flat --sim mujoco \
  env.domain_rand.randomize_gravity=true \
  'env.domain_rand.gravity_range=[[0.0,0.0,-10.5],[0.0,0.0,-8.5]]'

Common lifecycle boundaries:

Init-lifecycle items change model identity or geometry and must run during env/backend initialization.
Reset-lifecycle items perturb state or model parameters at reset through a backend-supported payload.
Interval-lifecycle items apply perturbations between steps.

The detailed task status and field semantics are in Domain Randomization.

Domain randomization is split by lifecycle: init, reset, and interval. The manager path is src/unilab/dr/manager.py; task providers live near the env owners, and backend capabilities are declared through src/unilab/base/backend/base.py.

Reset Gravity¶

Use --sim mujoco when enabling gravity reset randomization; Motrix does not advertise the same gravity capability in the current backend.

uv run train --algo ppo --task g1_walk_flat --sim mujoco \
  env.domain_rand.randomize_gravity=true \
  'env.domain_rand.gravity_range=[[0.0,0.0,-10.5],[0.0,0.0,-8.5]]'

Interval Push¶

uv run train --algo ppo --task g1_walk_flat --sim mujoco \
  env.domain_rand.push_robots=true \
  env.domain_rand.push_interval=500 \
  'env.domain_rand.max_force=[20.0,20.0,5.0]'

Owner-Local Defaults¶

Keep ranges in the task owner YAML when they are part of the task contract. For example, conf/ppo/task/go2_joystick_rough/mujoco.yaml enables base mass, center-of-mass, kp/kd, and push randomization, while conf/ppo/task/sharpa_inhand/mujoco.yaml configures object scale, friction, and force disturbance for Sharpa.

For the full current inventory, see Domain Randomization.