Remove duplicated episode_steps variable

scripts/config/environment/open_manipulator.py

@@ -2,6 +2,7 @@ from math import pi
 
 from geometry_msgs.msg import Quaternion
 
+
 config = {
     "ENV_NAME": "OpenManipulatorReacher",
     "TERM_COUNT": 10,
@@ -41,7 +42,6 @@ config = {
     "LOWER_Z": 0.116,
     "ENV_MODE": "sim",
     "TRAIN_MODE": True,
-    "MAX_EPISODE_STEPS": 100,
     "DISTANCE_THRESHOLD": 0.1,
     "REWARD_RESCALE_RATIO": 1.0,
     "REWARD_FUNC": "l2",

scripts/envs/open_manipulator/open_manipulator_reacher_env.py

@@ -33,7 +33,6 @@ class OpenManipulatorReacherEnv(gym.Env):
         self.cfg = cfg
         self.env_name = self.cfg["ENV_NAME"]
         self.env_mode = self.cfg["ENV_MODE"]
-        self._max_episode_steps = self.cfg["MAX_EPISODE_STEPS"]
         self.reward_rescale_ratio = self.cfg["REWARD_RESCALE_RATIO"]
         self.reward_func = self.cfg["REWARD_FUNC"]
 
@@ -43,7 +42,6 @@ class OpenManipulatorReacherEnv(gym.Env):
         else:
             self.ros_interface = OpenManipulatorRosRealInterface()
 
-        self.episode_steps = 0
         self.done = False
         self.reward = 0
 
@@ -67,27 +65,19 @@ class OpenManipulatorReacherEnv(gym.Env):
         Returns:
             Tuple of obs, reward_rescale * reward, done
         """
-        if action is None:
-            action = np.array([1, 1, 1, 1, 1, 1])
-
         self.done = False
 
-        if self.episode_steps == self._max_episode_steps:
-            self.done = True
-            self.episode_steps = 0
-
         act = action.flatten().tolist()
         self.ros_interface.set_joints_position(act)
 
         if self.env_mode == "sim":
             self.reward = self.compute_reward()
             if self.ros_interface.check_for_termination():
-                print ("Terminates current Episode : OUT OF BOUNDARY")
+                self.done = True
             elif self.ros_interface.check_for_success():
-                print ("Succeeded current Episode")
-        obs = self.ros_interface.get_observation()
+                self.done = True
 
-        self.episode_steps += 1
+        obs = self.ros_interface.get_observation()
 
         return obs, self.reward_rescale_ratio * self.reward, self.done, None
 

scripts/envs/open_manipulator/ros_interface.py

@@ -287,7 +287,7 @@ class OpenManipulatorRosBaseInterface(object):
         if dist < self.distance_threshold:
             self.success_count += 1
             if self.success_count == self.cfg["SUCCESS_COUNT"]:
-                self.done = True
+                print("Current episode succeeded")
                 self.success_count = 0
                 return True
             else:
@@ -344,7 +344,7 @@ class OpenManipulatorRosBaseInterface(object):
             rospy.logwarn("OUT OF BOUNDARY : joint_1_limit exceeds")
 
         if self.termination_count == term_count:
-            self.done = True
+            print("Current episode terminated")
             self.termination_count = 0
             return True
         else: