5 files changed, 11 insertions(+), 76 deletions(-)

M src/ddpg.py
M src/envs.py
M src/qlearning.py
M src/td3.py
M src/util.py

@@ 20,13 20,6 @@ from util import interrupt_handler
 from util import ReplayBuffer, ActionNormalizer
 from util import ActionNoise, NormalActionNoise, OUNoise
 
-import rospy
-from cv_bridge import CvBridge, CvBridgeError
-from sensor_msgs.msg import Image
-from gazebo_msgs.msg import ModelStates
-from geometry_msgs.msg import Twist
-from std_srvs.srv import Empty
-
 import signal
 import random
 from typing import List, Tuple

@@ 1,11 1,6 @@
-import signal
-import sys
 import time
 import gym
 import numpy as np
-import cv2 as cv
-
-
 
 import rospy
 from sensor_msgs.msg import Image


@@ 13,8 8,7 @@ from gazebo_msgs.msg import ModelStates
 from geometry_msgs.msg import Twist
 from std_srvs.srv import Empty
 
-
-from util import _process_image
+from util import process_image
 
 import gym
 import numpy as np


@@ 45,8 39,8 @@ class GazeboAutoVehicleEnv(gym.Env):
                                                np.array([1]),
                                                dtype=np.float32)
         else:
-            self.action_space = gym.spaces.Box(np.array([-1, 1.0]),
-                                               np.array([1, 1.5]),
+            self.action_space = gym.spaces.Box(np.array([-1, 0.3]),
+                                               np.array([1, 0.7]),
                                                dtype=np.float32)
 
         self.observation_space = gym.spaces.Box(np.array([-1]),


@@ 66,7 60,7 @@ class GazeboAutoVehicleEnv(gym.Env):
         self.state = None
 
     def image_callback(self, img):
-        self.state = _process_image(img, False)
+        self.state = process_image(img, False)
         pass
 
     def modelstate_callback(self, states):

@@ 2,13 2,10 @@ import gym
 import random
 import time
 import signal
-import sys
 
 import numpy as np
-import cv2 as cv
 
 import rospy
-from cv_bridge import CvBridge, CvBridgeError
 from sensor_msgs.msg import Image
 from gazebo_msgs.msg import ModelStates
 from geometry_msgs.msg import Twist


@@ 17,13 14,7 @@ from std_srvs.srv import Empty
 import numpy as np
 import matplotlib.pyplot as plt
 
-from util import _process_image, interuppt_handler
-
-
-def interuppt_handler(signum, frame):
-    print("Signal handler!!!")
-    sys.exit(-2) #Terminate process here as catching the signal removes the close process behaviour of Ctrl-C
-
+from util import process_image, interuppt_handler
 
 
 def simulate(env, q_table, alpha, epsilon, epsilon_decay):


@@ 124,7 115,7 @@ class GazeboAutoVehicleEnv():
         # self.unpause = rospy.ServiceProxy(self.GZUNPAUSE_TOPIC, Empty)
 
     def image_callback(self, img):
-        state = _process_image(img, True)
+        state = process_image(img, True)
         if state == None:
             self.state = None
             return

@@ 8,13 8,6 @@ from util import interrupt_handler
 from util import ReplayBuffer, ActionNormalizer
 from util import ActionNoise, NormalActionNoise, OUNoise
 
-import rospy
-from cv_bridge import CvBridge, CvBridgeError
-from sensor_msgs.msg import Image
-from gazebo_msgs.msg import ModelStates
-from geometry_msgs.msg import Twist
-from std_srvs.srv import Empty
-
 import signal
 import random
 from typing import List, Tuple


@@ 332,7 325,6 @@ class TD3Agent():
         actor_losses: List[float],
         critic_losses: List[float],
     ):
-
         fig, ax = plt.subplots(3, 1, figsize=(30, 15))
         ax[0].plot(scores)
         ax[0].plot(avgscores)


@@ 348,41 340,6 @@ class TD3Agent():
         print("Loading model . . .")
         self.actor.load_state_dict(torch.load('%s/%s_actor.pth' % (directory, filename)))
 
-"""## Environment
-*ActionNormalizer* is an action wrapper class to normalize the action values
-ranged in (-1. 1). Thanks to this class, we can make the agent simply select
-action values within the zero centered range (-1, 1).
-"""
-
-class ActionNormalizer(gym.ActionWrapper):
-    """Rescale and relocate the actions."""
-
-    def action(self, action: np.ndarray) -> np.ndarray:
-        """Change the range (-1, 1) to (low, high)."""
-        low = self.action_space.low
-        high = self.action_space.high
-
-        scale_factor = (high - low) / 2
-        reloc_factor = high - scale_factor
-
-        action = action * scale_factor + reloc_factor
-        action = np.clip(action, low, high)
-
-        return action
-
-    def reverse_action(self, action: np.ndarray) -> np.ndarray:
-        """Change the range (low, high) to (-1, 1)."""
-        low = self.action_space.low
-        high = self.action_space.high
-
-        scale_factor = (high - low) / 2
-        reloc_factor = high - scale_factor
-
-        action = (action - reloc_factor) / scale_factor
-        action = np.clip(action, -1.0, 1.0)
-
-        return action
-
 
 signal.signal(signal.SIGINT, interrupt_handler)
 env = GazeboAutoVehicleEnv(600, 800, 2, 1)


@@ 411,4 368,5 @@ agent.train(num_frames = 50000)
 # model_filename += "_7"
 # agent.load(directory="./saves", filename=model_filename)
 # while True:
+#     env.use_pause = False
 #     agent.test()

@@ 1,20 1,19 @@
-import signal
 import sys
+import random
 import numpy as np
 from typing import Dict
 import copy
 
 import gym
 import cv2 as cv
-from cv_bridge import CvBridge, CvBridgeError
+from cv_bridge import CvBridge
 
 
 def interrupt_handler(signum, frame):
-    print("Signal handler!!!")
+    print("Signal handler!!!", signum, frame)
     sys.exit(-2)
 
-
-def _process_image(img, show_image=True):
+def process_image(img, show_image=True):
     bridge = CvBridge()
     image = bridge.imgmsg_to_cv2(img, "bgr8")
     gray = bridge.imgmsg_to_cv2(img, "mono8")