Merge branch 'master' into imports

environments/carla_environment_wrapper.py

@@ -177,7 +177,7 @@ class CarlaEnvironmentWrapper(ew.EnvironmentWrapper):
                       - np.abs(self.control.steer) * 10
 
         # update measurements
-        self.observation = {
+        self.state = {
             'observation': sensor_data['CameraRGB'].data,
             'measurements': [measurements.player_measurements.forward_speed],
         }
@@ -221,9 +221,9 @@ class CarlaEnvironmentWrapper(ew.EnvironmentWrapper):
             self.game.start_episode(self.iterator_start_positions)
 
         # start the game with some initial speed
-        observation = None
+        state = None
         for i in range(self.num_speedup_steps):
-            observation = self.step([1.0, 0])['observation']
-        self.observation = observation
+            state = self.step([1.0, 0])['state']
+        self.state = state
 
-        return observation
+        return state

environments/doom_environment_wrapper.py

@@ -137,7 +137,7 @@ class DoomEnvironmentWrapper(ew.EnvironmentWrapper):
         # extract all data from the current state
         state = self.game.get_state()
         if state is not None and state.screen_buffer is not None:
-            self.observation = {
+            self.state = {
                 'observation': state.screen_buffer,
                 'measurements': state.game_variables,
             }

environments/gym_environment_wrapper.py

@@ -91,7 +91,7 @@ class GymEnvironmentWrapper(ew.EnvironmentWrapper):
             self.timestep_limit = self.env.spec.timestep_limit
         else:
             self.timestep_limit = None
-        self.measurements_size = len(self.step(0)['info'].keys())
+        self.measurements_size = (len(self.step(0)['info'].keys()),)
         self.random_initialization_steps = self.tp.env.random_initialization_steps
 
     def _wrap_state(self, state):