Merge pull request #9 from samholt:main

PiperOrigin-RevId: 718086218
Change-Id: I89e1489f67ae634e87ea082583841bee86fdb1c8

learning/train_jax_ppo.py

@@ -14,18 +14,12 @@
 # ==============================================================================
 """Train a PPO agent using JAX on the specified environment."""
 
-import os
-
-xla_flags = os.environ.get("XLA_FLAGS", "")
-xla_flags += " --xla_gpu_triton_gemm_any=True"
-os.environ["XLA_FLAGS"] = xla_flags
-os.environ["XLA_PYTHON_CLIENT_PREALLOCATE"] = "false"
-os.environ["MUJOCO_GL"] = "egl"
-
 from datetime import datetime
 import functools
 import json
+import os
 import time
+import warnings
 
 from absl import app
 from absl import flags
@@ -39,7 +33,6 @@
 import jax.numpy as jp
 import mediapy as media
 from ml_collections import config_dict
-from ml_collections import config_flags
 import mujoco
 from orbax import checkpoint as ocp
 from tensorboardX import SummaryWriter
@@ -52,11 +45,16 @@
 from mujoco_playground.config import locomotion_params
 from mujoco_playground.config import manipulation_params
 
+xla_flags = os.environ.get("XLA_FLAGS", "")
+xla_flags += " --xla_gpu_triton_gemm_any=True"
+os.environ["XLA_FLAGS"] = xla_flags
+os.environ["XLA_PYTHON_CLIENT_PREALLOCATE"] = "false"
+os.environ["MUJOCO_GL"] = "egl"
+
 # Ignore the info logs from brax
 logging.set_verbosity(logging.WARNING)
 
 # Suppress warnings
-import warnings
 
 # Suppress RuntimeWarnings from JAX
 warnings.filterwarnings("ignore", category=RuntimeWarning, module="jax")
@@ -267,11 +265,11 @@ def main(argv):
   print(f"Checkpoint path: {ckpt_path}")
 
   # Save environment configuration
-  with open(ckpt_path / "config.json", "w") as fp:
+  with open(ckpt_path / "config.json", "w", encoding="utf-8") as fp:
     json.dump(env_cfg.to_json(), fp, indent=4)
 
   # Define policy parameters function for saving checkpoints
-  def policy_params_fn(current_step, make_policy, params):
+  def policy_params_fn(current_step, make_policy, params):  # pylint: disable=unused-argument
     orbax_checkpointer = ocp.PyTreeCheckpointer()
     save_args = orbax_utils.save_args_from_target(params)
     path = ckpt_path / f"{current_step}"
@@ -352,7 +350,7 @@ def progress(num_steps, metrics):
   )
 
   # Train or load the model
-  make_inference_fn, params, _ = train_fn(
+  make_inference_fn, params, _ = train_fn(  # pylint: disable=no-value-for-parameter
       environment=env,
       progress_fn=progress,
       eval_env=None if _VISION.value else eval_env,
@@ -389,7 +387,7 @@ def progress(num_steps, metrics):
   rollout = [state0]
 
   # Run evaluation rollout
-  for i in range(env_cfg.episode_length):
+  for _ in range(env_cfg.episode_length):
     act_rng, rng = jax.random.split(rng)
     ctrl, _ = jit_inference_fn(state.obs, act_rng)
     state = jit_step(state, ctrl)

learning/train_rsl_rl.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
+# pylint: disable=wrong-import-position
 """Train a PPO agent using RSL-RL for the specified environment."""
 
 import os
@@ -28,7 +29,6 @@
 from absl import flags
 from absl import logging
 import jax
-import jax.numpy as jp
 import mediapy as media
 from ml_collections import config_dict
 import mujoco
@@ -136,7 +136,9 @@ def main(argv):
     wandb.config.update({"env_name": _ENV_NAME.value})
 
   # Save environment config to JSON
-  with open(os.path.join(ckpt_path, "config.json"), "w") as fp:
+  with open(
+      os.path.join(ckpt_path, "config.json"), "w", encoding="utf-8"
+  ) as fp:
     json.dump(env_cfg.to_json(), fp, indent=4)
 
   # Domain randomization
@@ -146,7 +148,7 @@ def main(argv):
   render_trajectory = []
 
   # Callback to gather states for rendering
-  def render_callback(env, state):
+  def render_callback(_, state):
     render_trajectory.append(state)
 
   # Create the environment
@@ -231,8 +233,11 @@ def render_callback(env, state):
   fps = 1.0 / base_env.dt / render_every
   traj = rollout[::render_every]
   frames = eval_env.render(
-      traj, camera=_CAMERA.value, height=480, width=640,
-      scene_option=scene_option
+      traj,
+      camera=_CAMERA.value,
+      height=480,
+      width=640,
+      scene_option=scene_option,
   )
   media.write_video("rollout.mp4", frames, fps=fps)
   print("Rollout video saved as 'rollout.mp4'.")

mujoco_playground/_src/locomotion/g1/randomize.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
+"""Utilities for randomization."""
 import jax
 from mujoco import mjx
 

mujoco_playground/_src/locomotion/locomotion_test.py

@@ -24,6 +24,7 @@
 
 
 class TestSuite(parameterized.TestCase):
+  """Tests for the locomotion environments."""
 
   @parameterized.named_parameters(
       {"testcase_name": f"test_can_create_{env_name}", "env_name": env_name}

mujoco_playground/_src/manipulation/franka_emika_panda/open_cabinet.py

@@ -19,8 +19,8 @@
 import jax
 import jax.numpy as jp
 from ml_collections import config_dict
-import mujoco
 from mujoco import mjx
+import mujoco  # pylint: disable=unused-import
 from mujoco.mjx._src import math
 
 from mujoco_playground._src import collision

mujoco_playground/_src/manipulation/franka_emika_panda/pick_cartesian.py

@@ -88,7 +88,7 @@ class PandaPickCubeCartesian(pick.PandaPickCube):
   """Environment for training the Franka Panda robot to pick up a cube in
   Cartesian space."""
 
-  def __init__(
+  def __init__(  # pylint: disable=non-parent-init-called,super-init-not-called
       self,
       config=default_config(),
       config_overrides: Optional[Dict[str, Union[str, int, list[Any]]]] = None,

mujoco_playground/_src/manipulation/franka_emika_panda/randomize_vision.py

@@ -60,9 +60,9 @@ def domain_randomize(
 ) -> Tuple[mjx.Model, mjx.Model]:
   """Tile the necessary axes for the Madrona BatchRenderer."""
   mj_model = pick_cartesian.PandaPickCubeCartesian().mj_model
-  FLOOR_GEOM_ID = mj_model.geom('floor').id
-  BOX_GEOM_ID = mj_model.geom('box').id
-  STRIP_GEOM_ID = mj_model.geom('init_space').id
+  floor_geom_id = mj_model.geom('floor').id
+  box_geom_id = mj_model.geom('box').id
+  strip_geom_id = mj_model.geom('init_space').id
 
   in_axes = jax.tree_util.tree_map(lambda x: None, mjx_model)
   in_axes = in_axes.tree_replace({
@@ -93,16 +93,16 @@ def rand(rng: jax.Array, light_position: jax.Array):
     rgba = jp.array(
         [jax.random.uniform(key_box, (), minval=0.5, maxval=1.0), 0.0, 0.0, 1.0]
     )
-    geom_rgba = mjx_model.geom_rgba.at[BOX_GEOM_ID].set(rgba)
+    geom_rgba = mjx_model.geom_rgba.at[box_geom_id].set(rgba)
 
     strip_white = jax.random.uniform(key_strip, (), minval=0.8, maxval=1.0)
-    geom_rgba = geom_rgba.at[STRIP_GEOM_ID].set(
+    geom_rgba = geom_rgba.at[strip_geom_id].set(
         jp.array([strip_white, strip_white, strip_white, 1.0])
     )
 
     # Sample a shade of gray
     gray_scale = jax.random.uniform(key_floor, (), minval=0.0, maxval=0.25)
-    geom_rgba = geom_rgba.at[FLOOR_GEOM_ID].set(
+    geom_rgba = geom_rgba.at[floor_geom_id].set(
         jp.array([gray_scale, gray_scale, gray_scale, 1.0])
     )
 
@@ -112,11 +112,11 @@ def rand(rng: jax.Array, light_position: jax.Array):
         jax.random.randint(key_matid, shape=(num_geoms,), minval=0, maxval=10)
         + mat_offset
     )
-    geom_matid = geom_matid.at[BOX_GEOM_ID].set(
+    geom_matid = geom_matid.at[box_geom_id].set(
         -2
     )  # Use the above randomized colors
-    geom_matid = geom_matid.at[FLOOR_GEOM_ID].set(-2)
-    geom_matid = geom_matid.at[STRIP_GEOM_ID].set(-2)
+    geom_matid = geom_matid.at[floor_geom_id].set(-2)
+    geom_matid = geom_matid.at[strip_geom_id].set(-2)
 
     #### Cameras ####
     key_pos, key_ori, key = jax.random.split(key, 3)
@@ -134,7 +134,7 @@ def rand(rng: jax.Array, light_position: jax.Array):
     assert (
         nlight == 1
     ), f'Sim2Real was trained with a single light source, got {nlight}'
-    key_lsha, key_ldir, key_ldct, key = jax.random.split(key, 4)
+    key_lsha, key_ldir, key = jax.random.split(key, 3)
 
     # Direction
     shine_at = jp.array([0.661, -0.001, 0.179])  # Gripper starting position

mujoco_playground/_src/manipulation/leap_hand/rotate_z.py

@@ -23,7 +23,6 @@
 import numpy as np
 
 from mujoco_playground._src import mjx_env
-from mujoco_playground._src import reward
 from mujoco_playground._src.manipulation.leap_hand import base as leap_hand_base
 from mujoco_playground._src.manipulation.leap_hand import leap_hand_constants as consts
 
@@ -145,7 +144,7 @@ def step(self, state: mjx_env.State, action: jax.Array) -> mjx_env.State:
     rewards = {
         k: v * self._config.reward_config.scales[k] for k, v in rewards.items()
     }
-    reward = sum(rewards.values()) * self.dt
+    reward = sum(rewards.values()) * self.dt  # pylint: disable=redefined-outer-name
 
     state.info["last_last_act"] = state.info["last_act"]
     state.info["last_act"] = action

mujoco_playground/_src/manipulation/manipulation_test.py

@@ -24,6 +24,7 @@
 
 
 class TestSuite(parameterized.TestCase):
+  """Tests for the manipulation environments."""
 
   @parameterized.named_parameters(
       {"testcase_name": f"test_can_create_{env_name}", "env_name": env_name}

mujoco_playground/_src/mjx_env.py

@@ -28,7 +28,6 @@
 import numpy as np
 import tqdm
 
-
 # Root path is used for loading XML strings directly using etils.epath.
 ROOT_PATH = epath.Path(__file__).parent
 # Base directory for external dependencies.