multi-agent-rlenv 3.3.6__py3-none-any.whl → 3.3.7__py3-none-any.whl

marlenv/__init__.py

@@ -62,7 +62,7 @@ print(env.extras_shape) # (1, )
 If you want to create a new environment, you can simply create a class that inherits from `MARLEnv`. If you want to create a wrapper around an existing `MARLEnv`, you probably want to subclass `RLEnvWrapper` which implements a default behaviour for every method.
 """
 
-__version__ = "3.3.6"
+__version__ = "3.3.7"
 
 from . import models
 from . import wrappers

marlenv/adapters/overcooked_adapter.py

@@ -2,6 +2,7 @@ import sys
 from dataclasses import dataclass
 from typing import Literal, Sequence
 from copy import deepcopy
+from time import time
 
 import cv2
 import numpy as np
@@ -17,8 +18,10 @@ from overcooked_ai_py.visualization.state_visualizer import StateVisualizer
 @dataclass
 class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
     horizon: int
+    reward_shaping: bool
 
-    def __init__(self, oenv: OvercookedEnv):
+    def __init__(self, oenv: OvercookedEnv, reward_shaping: bool = True):
+        self.reward_shaping = reward_shaping
         self._oenv = oenv
         assert isinstance(oenv.mdp, OvercookedGridworld)
         self._mdp = oenv.mdp
@@ -86,10 +89,12 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
     def step(self, actions: Sequence[int] | npt.NDArray[np.int32 | np.int64]) -> Step:
         actions = [Action.ALL_ACTIONS[a] for a in actions]
         _, reward, done, info = self._oenv.step(actions, display_phi=True)
+        if self.reward_shaping:
+            reward += sum(info["shaped_r_by_agent"])
         return Step(
             obs=self.get_observation(),
             state=self.get_state(),
-            reward=np.array([reward]),
+            reward=np.array([reward], dtype=np.float32),
             done=done,
             truncated=False,
             info=info,
@@ -185,6 +190,32 @@ class Overcooked(MARLEnv[Sequence[int] | npt.NDArray, DiscreteActionSpace]):
             "you_shall_not_pass",
         ],
         horizon: int = 400,
+        reward_shaping: bool = True,
     ):
         mdp = OvercookedGridworld.from_layout_name(layout)
-        return Overcooked(OvercookedEnv.from_mdp(mdp, horizon=horizon))
+        return Overcooked(OvercookedEnv.from_mdp(mdp, horizon=horizon), reward_shaping=reward_shaping)
+
+    @staticmethod
+    def from_grid(
+        grid: Sequence[Sequence[Literal["S", "P", "X", "O", "D", "T", "1", "2", " "] | str]],
+        horizon: int = 400,
+        reward_shaping: bool = True,
+    ):
+        """
+        Create an Overcooked environment from a grid layout where
+        - S is a serving location
+        - P is a cooking pot
+        - X is a counter
+        - O is an onion dispenser
+        - D is a dish dispenser
+        - T is a tomato dispenser
+        - 1 is a player 1 starting location
+        - 2 is a player 2 starting location
+        - ' ' is a walkable space
+        """
+        # It is necessary to add an explicit layout name because Overcooked saves some files under this
+        # name. By default the name is a concatenation of the grid elements, which may include characters
+        # such as white spaces, pipes ('|') and square brackets ('[' and ']') that are invalid Windows file paths.
+        layout_name = str(time())
+        mdp = OvercookedGridworld.from_grid(grid, base_layout_params={"layout_name": layout_name})
+        return Overcooked(OvercookedEnv.from_mdp(mdp, horizon=horizon), reward_shaping=reward_shaping)

{multi_agent_rlenv-3.3.6.dist-info → multi_agent_rlenv-3.3.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: multi-agent-rlenv
-Version: 3.3.6
+Version: 3.3.7
 Summary: A strongly typed Multi-Agent Reinforcement Learning framework
 Project-URL: repository, https://github.com/yamoling/multi-agent-rlenv
 Author-email: Yannick Molinghen <yannick.molinghen@ulb.be>

{multi_agent_rlenv-3.3.6.dist-info → multi_agent_rlenv-3.3.7.dist-info}/RECORD

@@ -1,4 +1,4 @@
-marlenv/__init__.py,sha256=iEVXbl4mQmey7P2uFdeKEYWEmZ8QxNS_f52jNdw4nZs,3741
+marlenv/__init__.py,sha256=u27-QdgKv_1k3uR0oCBN7wcX2jRPmpICbJz1SaZ-f-A,3741
 marlenv/env_builder.py,sha256=s_lQANqP3iNc8nmcr3CanRVsExnn9qh0ihh4lFr0c4c,5560
 marlenv/env_pool.py,sha256=R3WIrnQ5Zvff4HR1ecfkDmuO2zl7v1ywQ0K2_nvWFzs,1070
 marlenv/exceptions.py,sha256=gJUC_2rVAvOfK_ypVFc7Myh-pIfSU3To38VBVS_0rZA,1179
@@ -6,7 +6,7 @@ marlenv/mock_env.py,sha256=qB0fYFIfbopJf7Va8kCeVI5vsOy1-2JdEYe9gdV1Ruw,4761
 marlenv/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 marlenv/adapters/__init__.py,sha256=rWiqQOqTx3kVL5ZkPo3rkczrlQBBhQbU55zGI26SEeY,929
 marlenv/adapters/gym_adapter.py,sha256=6CBEjANViTJBTUBmtVyrhJrzjBJxNs_4hmMnXXG2mkU,2906
-marlenv/adapters/overcooked_adapter.py,sha256=JZhB50cQGWGjaHWuPwskUKr6YthEptpYC3cD7i9GVvk,6832
+marlenv/adapters/overcooked_adapter.py,sha256=Ehwwha_gh9wsQWBVLvwKYR_P6WUco-W2LoxumVjXSPQ,8289
 marlenv/adapters/pettingzoo_adapter.py,sha256=4F1au6uctsqRhGfcZOeDRH-8hmrFXnA5xH1Z1Pnek3s,2870
 marlenv/adapters/pymarl_adapter.py,sha256=x__E90XpFbfSWhnBHtkcD6WYkmKki1LByNbUFoDBUcg,3416
 marlenv/adapters/smac_adapter.py,sha256=fOfKo1hL4ioKtM5qQGcwtfdkdwUEACjAZqaGmkoQUcU,8373
@@ -31,7 +31,7 @@ marlenv/wrappers/penalty_wrapper.py,sha256=v4_H8OEN2-yujLzRb6P7W7KwmXHtjAFsxcdp3
 marlenv/wrappers/rlenv_wrapper.py,sha256=C2XekgBIM4x3Wa2Mtsn7rihRD4ymC2hORI473Af0sfw,2962
 marlenv/wrappers/time_limit.py,sha256=CDIMMJPMyIDHSFxUJaC7nb7Kd86-07NgZeFhrpZm82o,3985
 marlenv/wrappers/video_recorder.py,sha256=d5AFu6qHqby9mOcBsYWYPxAPiK1vtnfMYdZ81AnCekI,2624
-multi_agent_rlenv-3.3.6.dist-info/METADATA,sha256=oHsLxFw-wlgzPyswB6r3QIWZWM_injRIFJuJczyZDTo,4897
-multi_agent_rlenv-3.3.6.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-multi_agent_rlenv-3.3.6.dist-info/licenses/LICENSE,sha256=_eeiGVoIJ7kYt6l1zbIvSBQppTnw0mjnYk1lQ4FxEjE,1074
-multi_agent_rlenv-3.3.6.dist-info/RECORD,,
+multi_agent_rlenv-3.3.7.dist-info/METADATA,sha256=zAUWp4QbbEnxN7EVkJfJAXKYEjAcfUIKaogM43fAjR8,4897
+multi_agent_rlenv-3.3.7.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+multi_agent_rlenv-3.3.7.dist-info/licenses/LICENSE,sha256=_eeiGVoIJ7kYt6l1zbIvSBQppTnw0mjnYk1lQ4FxEjE,1074
+multi_agent_rlenv-3.3.7.dist-info/RECORD,,