Upgrade WrapperNormalizeActions to match Dict&Tuple action_spaces.

digideep/environment/wrappers.py

@@ -77,26 +77,37 @@ def observation(self, observation):
 
 class WrapperNormalizedActions(gym.ActionWrapper):
     """
-    This is taken from `RL-Adventure-2 <https://github.com/higgsfield/RL-Adventure-2>`__.
+    This is based on `RL-Adventure-2 <https://github.com/higgsfield/RL-Adventure-2>`__.
     """
     def action(self, action):
-        low  = self.action_space.low
-        high = self.action_space.high
-
-        action = low + (action + 1.0) * 0.5 * (high - low)
-        action = np.clip(action, low, high)
-
-        return action
+        # TODO: What if action is a Dict?
+        return self._get_action_direct(self.action_space, action, reversed=False)
 
     def reverse_action(self, action):
-        low  = self.action_space.low
-        high = self.action_space.high
-
-        action = 2 * (action - low) / (high - low) - 1
-        action = np.clip(action, low, high)
+        return self._get_action_direct(self.action_space, action, reversed=True)
+
+    def _get_action_direct(self, action_space, action, reversed = False):
+        if isinstance(action_space, spaces.Dict):
+            action_ret = {}
+            for key in action_space.spaces:
+                action_ret[key] = self._get_action_direct(action_space.spaces[key], action[key])
+        elif isinstance(action_space, spaces.Tuple):
+            action_ret = []
+            for index in range(len(action_space.spaces)):
+                action_ret += [self._get_action_direct(action_space.spaces[index], action[index])]
+            action_ret = tuple(action_ret)
+        else:
+            low  = action_space.low
+            high = action_space.high
 
-        return actions
-
+            if reversed:
+                action_ret = 2 * (action - low) / (high - low) - 1
+                action_ret = np.clip(action_ret, low, high)
+            else:
+                action_ret = low + (action + 1.0) * 0.5 * (high - low)
+                action_ret = np.clip(action_ret, low, high)
+
+        return action_ret
 
 class WrapperDummyMultiAgent(gym.ActionWrapper):
     """