Adapted LayoutConverter to not be integrated in gym_env.py but to be called...

Adapted LayoutConverter to not be integrated in gym_env.py but to be called seperately with a command line argument. Additionally, all layouts are now stored in the config/layouts folder and no layout files with a different format are stored within our project.

cooperative_cuisine/configs/layouts/overcooked-ai/1-cramped-room.layout

-##U##
-NA_AN
-#___#
-#P#$#
\ No newline at end of file

cooperative_cuisine/configs/layouts/overcooked-ai/2-asymmetric-advantages.layout

-#########
-N_#$#N#_$
-#_A_U_A_#
-#___U___#
-###P#P###
\ No newline at end of file

cooperative_cuisine/configs/layouts/overcooked-ai/3-coordination-ring.layout

-###U#
-#__AU
-P_#_#
-NA__#
-#N$##

cooperative_cuisine/configs/layouts/overcooked-ai/4-forced-coordination.layout

-###U#
-N_#AU
-N_#_#
-PA#_#
-###$#

cooperative_cuisine/configs/layouts/overcooked-ai/5-counter-circuit.layout

-###UU###
-#A_____#
-P_####_$
-#_____A#
-###NN###
\ No newline at end of file

cooperative_cuisine/reinforcement_learning/config/environment/environment_config_rl.yaml

@@ -11,7 +11,7 @@ game:
   undo_dispenser_pickup: true
   validate_recipes: false
 
-layout_name: rl_small.layout
+layout_name: configs/layouts/rl/rl_small.layout
 
 layout_chars:
   _: Free

cooperative_cuisine/reinforcement_learning/config/environment/environment_config_rl_deterministic_order_generation.yaml

@@ -11,7 +11,7 @@ game:
   undo_dispenser_pickup: true
   validate_recipes: false
 
-layout_name: rl_small.layout
+layout_name: configs/layouts/rl/rl_small.layout
 
 layout_chars:
   _: Free

cooperative_cuisine/reinforcement_learning/config/environment/environment_config_rl_small_rewards.yaml

@@ -11,7 +11,7 @@ game:
   undo_dispenser_pickup: true
   validate_recipes: false
 
-layout_name: rl_small.layout
+layout_name: configs/layouts/rl/rl_small.layout
 
 
 layout_chars:

cooperative_cuisine/reinforcement_learning/config/environment/overcooked-ai_environment_config.yaml

@@ -11,7 +11,7 @@ game:
   undo_dispenser_pickup: true
   validate_recipes: false
 
-layout_name: centre_pots.layout
+layout_name: configs/layouts/rl/rl_small.layout
 
 
 layout_chars:

cooperative_cuisine/reinforcement_learning/layouts/convert_overcooked_ai_layouts.py → cooperative_cuisine/reinforcement_learning/convert_overcooked_ai_layouts.py

-from pathlib import Path
+import argparse
+from pathlib import Path, PurePath
+from cooperative_cuisine import ROOT_DIR
 
 
-def convert_overcookd_ai_layouts(file_root, filename):
+def convert_overcookd_ai_layouts():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--file", dest="inputfile", help="Input file path", required=True
+    )
+    args = parser.parse_args()
+    filepath = PurePath(args.inputfile)
+    print(filepath)
     convertion_dict = {
         " ": "_",
         "X": "#",
@@ -11,20 +20,20 @@ def convert_overcookd_ai_layouts(file_root, filename):
         "D": "P",
         "S": "$",
         "1": "A",
-        "2": "_"
+        "2": "A"
     }
-    loadpath = Path(file_root) / "reinforcement_learning" / "layouts" / "overcooked_ai_layouts" / filename
-    savepath = Path(file_root) / "reinforcement_learning" / "layouts" / filename
-    with open(loadpath, "r") as f:
+
+    savepath = Path(ROOT_DIR) / "configs" / "layouts" / "overcooked-ai" / filepath.name
+    with open(args.inputfile, "r") as f:
         layoutfile = f.read()
         f.close()
-        layout = eval(layoutfile)
-        lines = layout["grid"].split("\n")
-        additional_info = []
-        for key in layout:
-            if key != "grid":
-                additional_info.append(
-                    '; {}: {}'.format(key, str(layout[key]).replace("'", "").replace("None", "null")))
+    layout = eval(layoutfile)
+    lines = layout["grid"].split("\n")
+    additional_info = []
+    for key in layout:
+        if key != "grid":
+            additional_info.append(
+                '; {}: {}'.format(key, str(layout[key]).replace("'", "").replace("None", "null")))
 
         with open(savepath, "w+") as f:
             for line in lines:
@@ -39,4 +48,4 @@ def convert_overcookd_ai_layouts(file_root, filename):
 
 
 if __name__ == "__main__":
-    convert_overcookd_ai_layouts(Path().resolve().parents[1], "cramped_corridor.layout")
+    convert_overcookd_ai_layouts()

cooperative_cuisine/reinforcement_learning/gym_env.py

@@ -12,7 +12,6 @@ import yaml
 from gymnasium import spaces, Env
 from hydra.utils import instantiate
 from omegaconf import OmegaConf
-from layouts.convert_overcooked_ai_layouts import convert_overcookd_ai_layouts
 
 from cooperative_cuisine import ROOT_DIR
 from cooperative_cuisine.action import ActionType, InterActionData, Action
@@ -35,13 +34,18 @@ class SimpleActionSpace(Enum):
 
 
 def get_env_action(player_id, simple_action, duration):
+
     """
-    Creates a concrete action.
-    :param player_id:id of the player
-    :param simple_action: an action in the form of a SimpleActionSpace
-    :param duration: for how long an action should be conducted
-    :return: a concrete action
+
+    Args:
+        player_id: id of the player
+        simple_action: an action in the form of a SimpleActionSpace
+        duration: for how long an action should be conducted
+
+    Returns: a concrete action
+
     """
+
     match simple_action:
         case SimpleActionSpace.Up:
             return Action(
@@ -93,9 +97,11 @@ with open(ROOT_DIR / "pygame_2d_vis" / "visualization.yaml", "r") as file:
 
 
 def shuffle_counters(env):
+
     """
     Shuffles the counters of an environment
-    :param env: the environment object
+    Args:
+        env: the environment object
     """
     sample_counter = []
     other_counters = []
@@ -135,8 +141,10 @@ class EnvGymWrapper(Env):
 
     def __init__(self, config):
         """
-        :param config:  gets the rl and environment configuration from hydra
         Initializes all necessary variables
+
+        Args:
+            config:gets the rl and environment configuration from hydra
         """
         super().__init__()
         self.gridsize = 40
@@ -151,10 +159,7 @@ class EnvGymWrapper(Env):
         self.config_env = config_env
         self.config_item_info = config_item_info
         layout_file = config_env["layout_name"]
-        layout_path: Path = ROOT_DIR / "reinforcement_learning" / "layouts" / layout_file
-
-        if not Path(layout_path).is_file():
-            convert_overcookd_ai_layouts(ROOT_DIR, layout_file)
+        layout_path: Path = ROOT_DIR / layout_file
         with open(layout_path, "r") as file:
             self.layout = file.read()
         self.env: Environment = Environment(

cooperative_cuisine/reinforcement_learning/layouts/overcooked_ai_layouts/1-cramped-room.layout

-##U##
-NA_AN
-#___#
-#P#$#
\ No newline at end of file

cooperative_cuisine/reinforcement_learning/layouts/overcooked_ai_layouts/centre_pots.layout

-{
-    "grid":  """XXXOSSX
-                X  1  X
-                X P P X
-                X  2  X
-                XDDOXXX""",
-    "start_bonus_orders": [],
-    "start_all_orders" : [
-        { "ingredients" : ["onion", "onion", "onion"]}
-    ],
-    "rew_shaping_params": None
-}

cooperative_cuisine/reinforcement_learning/layouts/overcooked_ai_layouts/cramped_corridor.layout

-{
-    "grid":  """XTPXXXXSX
-                D      2X
-                XXXX XXXX
-                D      1X
-                XOPXXXXSX""",
-    "start_bonus_orders": [
-        { "ingredients" : ["tomato", "onion"]},
-        { "ingredients" : ["tomato"]},
-        { "ingredients" : ["tomato", "tomato"]}
-    ],
-    "start_all_orders" : [
-        { "ingredients" : ["tomato"]},
-        { "ingredients" : ["onion", "onion", "onion"]},
-        { "ingredients" : ["tomato", "tomato", "tomato"]},
-        { "ingredients" : ["tomato", "onion"] },
-        { "ingredients" : ["tomato", "tomato"]}
-    ],
-    "onion_value" : 21,
-    "tomato_value" : 13,
-    "onion_time" : 15,
-    "tomato_time" : 7
-}
\ No newline at end of file

cooperative_cuisine/reinforcement_learning/layouts/rl_small.layout

-##X#
-T__L
-U__P
-#C$#

cooperative_cuisine/reinforcement_learning/obs_converter/base_converter.py

@@ -11,6 +11,7 @@ class BaseStateConverter(StateToObservationConverter):
     """
     Converts an environment state to an Encoding where each counter/item has its unique value
     """
+
     def __init__(self):
         self.onehot = False
         self.grid_height: int | None = None
@@ -65,6 +66,15 @@ class BaseStateConverter(StateToObservationConverter):
 
     def convert_state_to_observation(self, env) -> np.ndarray:
 
+        """
+        Convert the environment into an onehot encoding
+        Args:
+            env: The environment object used
+
+        Returns: An encoding for the environment state that is not onehot
+
+        """
+
         grid_base_array = np.zeros(
             (
                 self.grid_width,

cooperative_cuisine/reinforcement_learning/obs_converter/base_converter_onehot.py

@@ -11,6 +11,7 @@ class BaseStateConverterOnehot(StateToObservationConverter):
     """
     Converts an environment state to an Onehot Encoding
     """
+
     def __init__(self):
         """
         Constructor setting basic variables as attributes.
@@ -62,16 +63,25 @@ class BaseStateConverterOnehot(StateToObservationConverter):
     def setup(self, env):
         """
         Set the grid width and height according to the present environment
-        Args: Environment used
+
+        Args:
+            env: The environment object used
         """
+
         self.grid_width, self.grid_height = int(env.kitchen_width), int(
             env.kitchen_height)
 
     def convert_state_to_observation(self, env) -> np.ndarray:
+
         """
-        Converts the environment into an onehot encoding
-        Args: Environment
+        Convert the environment into an onehot encoding
+        Args:
+            env: The environment object used
+
+        Returns: An onehot encoding for the environment state
+
         """
+
         grid_base_array = np.zeros(
             (
                 self.grid_width,

cooperative_cuisine/reinforcement_learning/run_single_agent.py

 import time
+from pathlib import Path
 
 import cv2
 from stable_baselines3 import DQN, A2C, PPO
@@ -15,14 +16,13 @@ def main(cfg: DictConfig):
     loads the trained model and enables the user to see an example with the according rewards.
     """
     additional_config = OmegaConf.to_container(cfg.additional_configs, resolve=True)
-    model_save_path = additional_config["log_path"] + "/" + additional_config["checkpoint_path"] + "/" + \
-                      additional_config["project_name"] + "_" + OmegaConf.to_container(cfg.model, resolve=True)[
-                          "model_name"]
+    model_save_path = Path(additional_config["log_path"]) / Path(additional_config["checkpoint_path"]) / Path(
+        additional_config["project_name"] + "_" + OmegaConf.to_container(cfg.model, resolve=True)["model_name"])
     model_class = call(cfg.model.model_type_inference)
     model = model_class(model_save_path)
     env = EnvGymWrapper(cfg)
 
-    #check_env(env)
+    # check_env(env)
     obs, info = env.reset()
     print(obs)
     while True: