Update gym to gymnasium

algorithms/utils/act.py

@@ -1,8 +1,9 @@
+import gymnasium as gym
 import torch
 import torch.nn as nn
-import gym.spaces
-from .mlp import MLPLayer
+
 from .distributions import BetaShootBernoulli, Categorical, DiagGaussian, Bernoulli
+from .mlp import MLPLayer
 
 
 class ACTLayer(nn.Module):

algorithms/utils/flatten.py

@@ -1,15 +1,16 @@
-import gym.spaces
+import gymnasium.spaces
 import numpy as np
+import gymnasium as gym
 from collections import OrderedDict
 
 
 def build_flattener(space):
-    if isinstance(space, gym.spaces.Dict):
+    if isinstance(space, gymnasium.spaces.Dict):
         return DictFlattener(space)
-    elif isinstance(space, gym.spaces.Box) \
-            or isinstance(space, gym.spaces.MultiDiscrete):
+    elif isinstance(space, gymnasium.spaces.Box) \
+            or isinstance(space, gymnasium.spaces.MultiDiscrete):
         return BoxFlattener(space)
-    elif isinstance(space, gym.spaces.Discrete):
+    elif isinstance(space, gymnasium.spaces.Discrete):
         return DiscreteFlattener(space)
     else:
         raise NotImplementedError
@@ -21,15 +22,15 @@ class DictFlattener():
 
     def __init__(self, ori_space):
         self.space = ori_space
-        assert isinstance(ori_space, gym.spaces.Dict)
+        assert isinstance(ori_space, gymnasium.spaces.Dict)
         self.size = 0
         self.flatteners = OrderedDict()
         for name, space in self.space.spaces.items():
-            if isinstance(space, gym.spaces.Box):
+            if isinstance(space, gymnasium.spaces.Box):
                 flattener = BoxFlattener(space)
-            elif isinstance(space, gym.spaces.Discrete):
+            elif isinstance(space, gymnasium.spaces.Discrete):
                 flattener = DiscreteFlattener(space)
-            elif isinstance(space, gym.spaces.Dict):
+            elif isinstance(space, gymnasium.spaces.Dict):
                 flattener = DictFlattener(space)
             self.flatteners[name] = flattener
             self.size += flattener.size

algorithms/utils/utils.py

@@ -1,6 +1,7 @@
 import copy
 import math
-import gym.spaces
+
+import gymnasium as gym
 import numpy as np
 import torch
 import torch.nn as nn

envs/JSBSim/core/catalog.py

@@ -3,7 +3,7 @@
 from enum import Enum
 from collections import namedtuple
 from numpy.linalg import norm
-from gym.spaces import Box, Discrete
+from gymnasium.spaces import Box, Discrete
 from ..utils.utils import in_range_deg
 
 """
@@ -549,7 +549,10 @@ def add_jsbsim_props(self, jsbsim_props):
         Args:
             jsbsim_props (list): list of 'name_jsbsim (access)' of jsbsim properties
         """
-        for jsbsim_prop in jsbsim_props:
+        jsbsim_props_tmp=jsbsim_props.split("\n")
+        for jsbsim_prop in jsbsim_props_tmp:
+            if jsbsim_prop.strip() == "":
+                continue  # skip empty line
             [name_jsbsim, access] = jsbsim_prop.split(" ")
             access = re.sub(r"[\(\)]", "", access)  # remove parenthesis from the flag
             name = re.sub(r"_$", "", re.sub(r"[\-/\]\[]+", "_", name_jsbsim))  # get property name from jsbsim name

envs/JSBSim/envs/env_base.py

@@ -1,13 +1,13 @@
-import gym
-from gym.utils import seeding
+import gymnasium
+from gymnasium.utils import seeding
 import numpy as np
 from typing import Dict, Any, Tuple
 from ..core.simulatior import AircraftSimulator, BaseSimulator
 from ..tasks.task_base import BaseTask
 from ..utils.utils import parse_config
 
 
-class BaseEnv(gym.Env):
+class BaseEnv(gymnasium.Env):
     """
     A class wrapping the JSBSim flight dynamics module (FDM) for simulating
     aircraft as an RL environment conforming to the OpenAI Gym Env
@@ -35,11 +35,11 @@ def num_agents(self) -> int:
         return self.task.num_agents
 
     @property
-    def observation_space(self) -> gym.Space:
+    def observation_space(self) -> gymnasium.Space:
         return self.task.observation_space
 
     @property
-    def action_space(self) -> gym.Space:
+    def action_space(self) -> gymnasium.Space:
         return self.task.action_space
 
     @property

envs/JSBSim/tasks/heading_task.py

@@ -1,5 +1,5 @@
 import numpy as np
-from gym import spaces
+from gymnasium import spaces
 from .task_base import BaseTask
 from ..core.catalog import Catalog as c
 from ..reward_functions import AltitudeReward, HeadingReward

envs/JSBSim/tasks/multiplecombat_task.py

@@ -1,5 +1,5 @@
 import numpy as np
-from gym import spaces
+from gymnasium import spaces
 from typing import Tuple
 import torch
 

envs/JSBSim/tasks/singlecombat_task.py

@@ -1,6 +1,6 @@
 import torch
 import numpy as np
-from gym import spaces
+from gymnasium import spaces
 from typing import Literal
 from .task_base import BaseTask
 from ..core.simulatior import AircraftSimulator

envs/JSBSim/tasks/singlecombat_with_missle_task.py

@@ -1,5 +1,5 @@
 import numpy as np
-from gym import spaces
+from gymnasium import spaces
 from collections import deque
 
 from .singlecombat_task import SingleCombatTask, HierarchicalSingleCombatTask

envs/JSBSim/tasks/task_base.py

@@ -1,5 +1,5 @@
 import numpy as np
-from gym import spaces
+from gymnasium import spaces
 from typing import List, Tuple
 from abc import ABC, abstractmethod
 from ..core.catalog import Catalog as c

envs/JSBSim/test/test_baseline_use_obs.py

@@ -150,7 +150,7 @@ def test_maneuver():
             print(info)
             break
         step += 1
-    # plt.plot(reward_list)
+    plt.plot(reward_list)
     # plt.savefig('rewards.png')
 
 

runner/share_jsbsim_runner.py

@@ -1,12 +1,12 @@
-import time
-from matplotlib.pyplot import axis
-import torch
 import logging
-import numpy as np
-from gym import spaces
+import time
 from typing import List
-from .base_runner import Runner
+
+import numpy as np
+import torch
+
 from algorithms.utils.buffer import SharedReplayBuffer
+from .base_runner import Runner
 
 
 def _t2n(x):

scripts/train/train_gym.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 import sys
 import os
-import gym
+import gymnasium as gym
 import wandb
 import socket
 import torch
@@ -150,7 +150,7 @@ def main(args):
     if all_args.use_wandb:
         run = wandb.init(config=all_args,
                          project=all_args.env_name,
-                         entity=all_args.wandb_name,
+                         # entity=all_args.wandb_name,
                          notes=socket.gethostname(),
                          name=f"{all_args.experiment_name}_seed{all_args.seed}",
                          group=all_args.scenario_name,

scripts/train/train_jsbsim.py

@@ -111,7 +111,6 @@ def main(args):
     if all_args.use_wandb:
         run = wandb.init(config=all_args,
                          project=all_args.env_name,
-                         entity=all_args.wandb_name,
                          notes=socket.gethostname(),
                          name=f"{all_args.experiment_name}_seed{all_args.seed}",
                          group=all_args.scenario_name,

scripts/train_selfplay_shoot.sh

@@ -1,4 +1,5 @@
 #!/bin/sh
+
 env="SingleCombat"
 scenario="1v1/ShootMissile/HierarchySelfplay"
 algo="ppo"

scripts/train_share_selfplay.sh

@@ -1,4 +1,5 @@
 #!/bin/sh
+
 env="MultipleCombat"
 scenario="2v2/NoWeapon/HierarchySelfplay"
 algo="mappo"

tests/test_ppo.py

@@ -1,10 +1,11 @@
-import sys
 import os
-import torch
-import pytest
-import numpy as np
+import sys
 from itertools import product
-import gym.spaces
+
+import gymnasium as gym
+import numpy as np
+import pytest
+import torch
 
 sys.path.append(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))