def runApeX():
config = read_config("./Configs/configApeX.yml")
epsilons = config["min_epsilons"]
N_ACTORS = config['n_actors']
BaseManager.register('ReplayBM', Memory)
BaseManager.register('LearnerBM', Learner)
BaseManager.register('ServerBM', Server)
manager = BaseManager()
manager.start()
server = manager.ServerBM()
replay = manager.ReplayBM(config)
learner = manager.LearnerBM(config)
processes = [
Process(target=run,
args=(replay, learner, server, config, epsilons[p], p))
for p in range(N_ACTORS)
]
p_learner = Process(target=learner.train_nn, args=(
replay,
server,
))
p_learner.start()
for p in processes:
p.start()
p_learner.join()
for p in processes:
p.join()
import numpy as np
import torch
import torch.nn as nn
import torch.optim as optim
from collections import deque
import random
import time
import sys
from Models.Networks.SimpleNetwork import Network
from utils.tool import read_config, read_seed
np.random.seed(read_seed())
config = read_config("./Configs/configD2QN.yml")
N_ACTIONS = config["n_actions"]
GAMMA = config["gamma"]
EPSILON_DECAY = config["epsilon_decay"]
EPSILON_MIN = config["epsilon_min"]
BATCH_SIZE = config["batch_size"]
MIN_REPLAY_SIZE = config["min_replay_size"]
TARGET_UPDATE = config["target_update"]
SAMPLING_TIME = config["sampling_time"]
NAME = config["model_name"]
# Deep Q Network agent
class Agent:
def __init__(self):
import numpy as np
import torch
import torch.nn as nn
import torch.optim as optim
from collections import deque
import random
import time
import sys
from Models.Networks.NoisyDuelingNetwork import Network
from utils.tool import read_config, read_seed
np.random.seed(read_seed())
config = read_config("./Configs/configNoisy.yml")
N_ACTIONS = config["n_actions"]
GAMMA = config["gamma"]
EPSILON_DECAY = config["epsilon_decay"]
EPSILON_MIN = config["epsilon_min"]
BATCH_SIZE = config["batch_size"]
MIN_REPLAY_SIZE = config["min_replay_size"]
TARGET_UPDATE = config["target_update"]
SAMPLING_TIME = config["sampling_time"]
NAME = config["model_name"]
# Deep Q Network agent
class Agent:
def __init__(self):
#Initialisationn of environnment variables
from Models.DistributedModels.Apex.Main import runApeX
from utils.tool import read_config, read_seed, write_seed
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description='choose your model')
parser.add_argument('-model', type=str, default='DQN', help='DQN/D2QN/D3QN/Noisy/PER/ApeX')
parser.add_argument('-seed', type=int, default=-1, help='save/init')
args = parser.parse_args()
#seeds
if args.seed == -1:
config = read_config("./Configs/config" + args.model + ".yml")
write_seed(config["seed"])
else:
write_seed(args.seed)
print("seed:", read_seed())
#Models
if args.model == "ApeX":
runApeX()
elif args.model == "DQN":
env = EnvCarla.CarEnv(0)
agent = ModelDQN.Agent()
TrainModel.run(config, agent, env)
elif args.model == "D2QN":
env = EnvCarla.CarEnv(0)
(sys.version_info.major, sys.version_info.minor,
'win-amd64' if os.name == 'nt' else 'linux-x86_64'))[0])
except IndexError:
pass
import carla
import random
import time
import numpy as np
import cv2
import math
from utils.tool import read_config, rgb_to_gray
import pathlib
config = read_config("./Configs/confEnvironmentTest.yml")
SECONDS_PER_EPISODE = config['TIME_WP']
#Image's size
IMG_WIDTH = config['IMG_WIDTH']
IMG_HEIGHT = config['IMG_HEIGHT']
SHOW_CAM = config['SHOW_PREVIEW']
#Distance between two waypoints
NAV_I = config['NAV_I']
START_POSITION = config['position']
TRAJECTORY = config['trajectory']
#Norme
VMAX = config['VMAX']
import torch
import numpy as np
import time
import random, sys
from Models.Networks.NoisyDuelingNetwork import Network
import Environments.Environment as EnvCarla
from torch.utils.tensorboard import SummaryWriter
from utils.tool import read_config, read_seed
config = read_config("./Configs/configApeX.yml")
N_ACTIONS = config["n_actions"]
EPSILON_DECAY = config["epsilon_decay"]
UPDATE_REPLAY = config['update_replay']
UPDATE_NETWORK_TIME = config['update_network']
np.random.seed(read_seed())
class Actor:
def __init__ (self, config, num, epsilon_min):
#Define env
self.epsilon = config['epsilon']
self.epsilon_min = epsilon_min
self.local_replay = []
self.num = num
import numpy as np
import torch
import torch.nn as nn
import torch.optim as optim
from collections import deque
import random
import time
import sys
from Models.Networks.NoisyDuelingNetwork import Network
from Models.NoDistributedModels import ReplayBuffer
from utils.tool import read_config, read_seed
np.random.seed(read_seed())
config = read_config("./Configs/configPER.yml")
N_ACTIONS = config["n_actions"]
GAMMA = config["gamma"]
EPSILON_DECAY = config["epsilon_decay"]
EPSILON_MIN = config["epsilon_min"]
BATCH_SIZE = config["batch_size"]
MIN_REPLAY_SIZE = config["min_replay_size"]
TARGET_UPDATE = config["target_update"]
SAMPLING_TIME = config["sampling_time"]
NAME = config["model_name"]
# Double Deep Q Network agent
class Agent: