if __name__ == '__main__':
global world
rospy.init_node('ddpg')
pub_result = rospy.Publisher('result', String, queue_size=5)
ep_0 = rospy.get_param('~ep_number')
world = rospy.get_param('~file_path')
MAX_STEPS = rospy.get_param('~max_steps')
# MAX_STEPS = 50
if (ep_0 != 0):
trainer.load_models(ep_0)
rospy.loginfo("Starting at episode: %s ", str(ep_0))
result = Float32()
env = Env(action_dim=ACTION_DIMENSION)
before_training = 4
past_action = np.zeros(ACTION_DIMENSION)
for ep in range(ep_0, MAX_EPISODES):
done = False
state = env.reset()
if is_training and not ep % 10 == 0 and len(
replay_buffer) >= before_training * BATCH_SIZE:
rospy.loginfo("---------------------------------")
rospy.loginfo("Episode: %s training", str(ep))
rospy.loginfo("---------------------------------")
else:
if len(replay_buffer) >= before_training * BATCH_SIZE:
rospy.loginfo("---------------------------------")
rospy.loginfo("Episode: %s evaluating", str(ep))
if __name__ == '__main__':
global world
rospy.init_node('sac')
pub_result = rospy.Publisher('result', String, queue_size=5)
ep_0 = rospy.get_param('~ep_number')
world = rospy.get_param('~file_path')
max_steps = rospy.get_param('~max_steps')
if (ep_0 != 0):
agent.load_models(ep_0)
rospy.loginfo("Starting at episode: %s ", str(ep_0))
result = Float32()
env = Env()
before_training = 4
past_action = np.array([0.,0.])
for ep in range(ep_0, max_episodes):
done = False
state = env.reset()
if is_training and not ep%10 == 0 and len(replay_buffer) > before_training*batch_size:
rospy.loginfo("---------------------------------")
rospy.loginfo("Episode: %s training", str(ep))
rospy.loginfo("---------------------------------")
else:
if len(replay_buffer) > before_training*batch_size:
rospy.loginfo("---------------------------------")
rospy.loginfo("Episode: %s evaluating", str(ep))
rospy.loginfo("---------------------------------")