def run_game():
# 初始化游戏并创建一个屏幕对象
all_setting = AllSettings()
stats = Gamestats(all_setting)
pygame.init() # 初始化pygame
# 创建屏幕
screen = pygame.display.set_mode((all_setting.screen_width, all_setting.screen_height))
pygame.display.set_caption("Alien Invation")
# 创建飞船对象
ship = Ship(screen, all_setting)
# 创建一个用于存储子弹，外星人，外星人子弹的编组
bullets = Group()
aliens = Group()
enemy_bullets = Group()
shield = Shield(ship, screen)
# gf.creat_fleet(all_setting, screen, ship, aliens)
play_button = Button(all_setting, screen, "Play")
score_get = Scoreboard(all_setting, screen, stats)
# 开始游戏主循环
while True:
gf.check_events(all_setting, screen, ship, aliens, bullets, stats, play_button, score_get, enemy_bullets)
if stats.game_active:
ship.update()
shield.set_pos()
gf.update_bullets(bullets, aliens, all_setting, screen, ship, stats, score_get, enemy_bullets)
gf.update_aliens(all_setting, aliens, ship, stats, screen, bullets, score_get, shield, enemy_bullets)
gf.update_enemy_bullets(enemy_bullets, all_setting)
gf.update_screen(all_setting, screen, ship, aliens, bullets, play_button, stats, score_get, shield,
enemy_bullets)
def createShields(self):
self.tb.sorting('shit')
name = self.ui.lineEdit_shield_template.text()
prfx = self.ui.lineEdit_shield_prefix.text()
ot_x = self.ui.spinBox_ots_shit_X.value()
ot_y = self.ui.spinBox_ots_shit_Y.value()
list = self.tb.getTable('shit')
shit = Shield(name, self.tf, self.mf, ot_x, ot_y, prfx)
shit.createProgs(list)
self.statusBar().showMessage('Программы Щитов созданы!')
mes = Mess()
mes.MesProgComplete('Программы готовы', 'Программы Щитов созданы!')
def post(self):
methods = self.request.get('core')
url = self.request.get('url')
pers = self.request.get('custom_hash')
remoteip = self.request.remote_addr
"""
challenge = self.request.get('recaptcha_challenge_field')
response = self.request.get('recaptcha_response_field')
remoteip = environ['REMOTE_ADDR']
cResponse = captcha.submit(challenge,response,"6LfAO78SAAAAANHiQxEObRo36llHrkQa3kauMaD3",remoteip)
list_url="Errore"
"""
#if cResponse.is_valid:
shield = Shield(remoteip)
if shield.getShieldSentence()==False:
try :
if url[0:7]!="http://":
url ="http://"+url
#result = urlfetch.fetch(url=url,method=urlfetch.GET,follow_redirects=False)
#if result.status_code == 200:
if methods=="classic":
res = formats.DecodeJson(core.short(url))
list_url = res.getMessage()
elif methods=="custom":
res = formats.DecodeJson(core.short(url,pers))
list_url = res.getMessage()
else :
res = "non trovato"
list_url = res
#elif result.status_code in (300,301,302,303,307):
# res = "redirect"
# list_url = res
except :
res = "Indirizzo non valido"
list_url = res
else:
#list_url = cResponse.error_code
list_url = "Errore Numero tentativi orari superati"
template_values = {
'urls': list_url,
}
path = os.path.join(os.path.dirname(__file__), 'templates/result.html')
self.response.out.write(template.render(path, template_values))
def load(self, file_name):
with open(file_name, 'r') as read_json:
the_entire_file = read_json.read()
the_entire_dict = json.loads(the_entire_file)
for shield_name in the_entire_dict['shields']:
the_shield_dict = the_entire_dict['shields'][shield_name]
new_shield = Shield(shield_name, the_shield_dict['type'],
the_shield_dict['size'],
the_shield_dict['weight'],
the_shield_dict['strength'])
#prevents double loads, same name overwriting
if shield_name not in self.shields:
# shield_name is the string, new_shield is the Shield object!
self.shields[shield_name] = new_shield
for weapon_name in the_entire_dict['weapons']:
the_weapon_dict = the_entire_dict['weapons'][weapon_name]
new_weapon = Weapon(weapon_name, the_weapon_dict['type'],
the_weapon_dict['reach'],
the_weapon_dict['weight'],
the_weapon_dict['strength'])
#prevents double loads, same name overwriting
if weapon_name not in self.weapons:
# shield_name is the string, new_shield is the Shield object!
self.weapons[weapon_name] = new_weapon
print(self.weapons)
print(self.shields)
def __init__(self):
## sets dimension of the screen and the background
pygame.init()
self.w = 900
self.h = 600
self.screen = pygame.display.set_mode((self.w, self.h), pygame.FULLSCREEN)
self.background = pygame.image.load("background.png")
## creates all the objects and lists needed for the game
self.player = Player(self.w, self.h, self.background)
self.shooter = Shooter(self.player.x, self.player.y, self.player.d)
self.shield = Shield(self.player.x, self.player.y, self.player.d)
self.generator = Generator(self.player.x, self.player.y, self.player.d)
self.bullet = Bullet(self.shooter.x, self.shooter.y, self.shooter.image_rotation)
self.enemies = [Enemy(self.w, self.h)]
self.enemies_2 = []
self.counter_enemies_2_dead = 0
self.collectible = Collectible(self.w, self.h)
self.ray = Ray(self.w, self.h)
## loads energy image and sets its default value
self.energy_5 = pygame.image.load("energy_5.png")
self.energy = 100
## sets all default values
self.points = 0
self.killed_big_enemies = 0
self.killed_small_enemies = 0
self.rays_collected = 0
self.collectibles_collected = 0
## initializes fonts and creates the first text
font = pygame.font.SysFont("comic", 64)
self.font = pygame.font.SysFont("arial", 10)
text = font.render("Click to play", True, (255, 255, 255))
## loads sound
self.sound = pygame.mixer.Sound("blip.wav")
## sets timer and default timer variables
self.clock = pygame.time.Clock()
self.timer = 0
self.seconds = 0
self.minutes = 0
## manages beginning screen and first inputs
beginning = True
while beginning:
self.screen.fill((0, 0, 0))
self.screen.blit(text, ((self.w / 2) - 150, (self.h / 2) - 64))
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.display.quit()
if event.type == pygame.MOUSEBUTTONUP:
beginning = False
self.screen.blit(self.background, (0, 0))
self.screen.blit(self.player.images[1], (self.player.x, self.player.y))
self.screen.blit(self.shooter.image, (self.shooter.x, self.shooter.y))
self.running()
pygame.display.update()
def generate5(score):
dx = random.random()
if dx < 0.9: dx -= 1.0
dy = random.random()
if dy < 0.9: dy -= 1.0
mag = 1 + score / 100
dx *= mag
dy *= mag
side = random.randint(1, 4)
if side == 1: # left
x = 0
y = random.uniform(0, get_canvas_height())
if dx < 0: dx = -dx
elif side == 2: # bottom
x = random.uniform(0, get_canvas_width())
y = 0
if dy < 0: dy = -dy
elif side == 3: # right
x = get_canvas_width()
y = random.uniform(0, get_canvas_height())
if dx > 0: dx = -dx
else: # side == 4: # up
x = random.uniform(0, get_canvas_width())
y = get_canvas_height()
if dy > 0: dy = -dy
g = Shield((x, y), (dx, dy))
gfw.world.add(gfw.layer.shield, g)
def load_shields(self):
sprites = SpriteSheet("shields").load_strip((0, 0, 87, 84), 4)
return [
Shield([sprites[i]],
(.20 * (i + 1) * SCREEN_WIDTH, SCREEN_HEIGHT * 0.75))
for i in range(len(sprites))
]
def testShip():
alf = Shield("Standard Shield", 500, 12)
w1 = Weapon("Standard Issue Phaser Array", 132, "Directed Energy")
w2 = Weapon("Standard Issue Photon Torpedoes", 179, "Antimatter")
ship = Ship("VOY", "UFP", 2000, alf, [w1, w2], 200)
ship.addToCargoHold(ship.weapons[0])
ship.addToCargoHold(ship.weapons[1])
return ship
def _initEmptyBoard(self):
"""
:return: the initialized board
"""
board = [[Square(None, i, j) for j in range(5)] for i in range(5)]
board[0][0] = Square(Note(False), 0, 0)
board[0][1] = Square(Governance(False), 0, 1)
board[0][2] = Square(Relay(False), 0, 2)
board[0][3] = Square(Shield(False), 0, 3)
board[0][4] = Square(Drive(False), 0, 4)
board[1][4] = Square(Preview(False), 1, 4)
board[3][0] = Square(Preview(True), 3, 0)
board[4][0] = Square(Drive(True), 4, 0)
board[4][1] = Square(Shield(True), 4, 1)
board[4][2] = Square(Relay(True), 4, 2)
board[4][3] = Square(Governance(True), 4, 3)
board[4][4] = Square(Note(True), 4, 4)
return board
def __init__(self, starting_position, universe):
self.position_ = starting_position
self.universe_ = universe
self.font_size_ = 22
self.id_ = self.create_ID()
self.max_health_ = 100
self.selected_weapon_index_ = 0
self.shield_ = Shield(self.position_, 100)
self.size_ = 20
self.text_antialias_ = 1
self.ui_font_ = pygame.font.SysFont("monospace", self.font_size_)
self.weapon_label_x_spacing_ = 5
self.weapon_label_y_spacing_ = 10
self.weapons_ = [Rifle(self.universe_), HomingMissiles(self.universe_)]
self.current_weapon_ = self.weapons_[self.selected_weapon_index_]
self.health_ = self.max_health_
def game():
pygame.init()
play_intro()
grid = loadtxt('maze.txt', dtype=str)
rows, cols = grid.shape
width, height = (24, 24)
screen_size = (width*cols, height*rows)
screen = pygame.display.set_mode(screen_size, 0, 32)
block_list = []
for col in range(cols):
for row in range(rows):
value = grid[row][col]
if value == 'X':
pos = (col*width, row*height)
block_list.append(Brick((width, height), pos))
elif value == 'S':
pos = (col*width, row*height)
block_list.append(Shield((width, height), pos))
elif value == '-' or value == '|' or value == 'N':
pos = (col * width, row * height)
block_list.append(Pill((width, height), pos))
elif value == 'P':
pos = (col * width, row * height)
block_list.append(Powerpill((width, height), pos))
elif value == 'Y':
pos = (col * width, row * height)
block_list.append(Pacman((width, height), pos))
elif value == 'b':
pos = (col * width, row * height)
block_list.append(Blinky((width, height), pos))
elif value == 'i':
pos = (col * width, row * height)
block_list.append(Inky((width, height), pos))
elif value == 'p':
pos = (col * width, row * height)
block_list.append(Pinky((width, height), pos))
elif value == 'c':
pos = (col * width, row * height)
block_list.append(Clyde((width, height), pos))
background = pygame.Surface(screen_size).convert()
background.fill((0, 0, 0))
while True:
for event in pygame.event.get():
if event.type == QUIT:
exit()
# pacman.move()
# for tile in tiles:
# pacman.collide(tile)
screen.blit(background, (0, 0))
for blocks in block_list:
blocks.draw(screen)
# pacman.draw(screen)
pygame.display.update()
def __init__(self, x=0, y=0, vx=0, vy=0):
super().__init__(x, y, vx, vy)
self.oldVel = self.vel
self.input = {}
self.input['left'] = False
self.input['right'] = False
self.input['up'] = False
self.input['down'] = False
self.input['shot'] = False
self.input['dash'] = False
self.wasMoving = False
self.active = True
self.bullets = []
self.radius = Player.RADIUS
self.cooldownS = 0 #Player.SHOT_COOLDOWN
self.dashCool = 0 #Player.DASH_COOLDOWN
self.mouseDir = Vector(1, 0)
self.attackMode = Player.AttackMode.SHOOTING
self.shield = Shield(self.pos, self.radius * 1.2)
self.melee = Melee(self.pos, self.radius * 1.8)
def create_shield(self):
shield_name = input(
'\tWhat is the name of the shield we are creating? ')
shield_type = input(
'\tWhat is the type of the shield we are creating? ')
size = int(input('\tWhat is the shield size? '))
weight = int(input('\tWhat is the shield weight? '))
strength = int(input('\tWhat is the shield strength? '))
if shield_name not in self.shields:
self.shields[shield_name] = Shield(shield_name, shield_type, size,
weight, strength)
print('\t The shield %s has been created.' % shield_name)
def dcmotor (learning_method, number_of_rollouts, simulation_steps, learning_eposides, critic_structure, actor_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100):
A = np.matrix([[0.98965, 1.4747e-08], [7.4506e-09, 0]])
B = np.matrix([[128], [0]])
#intial state space
s_min = np.array([[-1.0], [-1.0]])
s_max = np.array([[1.0], [1.0]])
#sample an initial condition for system
x0 = np.matrix([
[random.uniform(s_min[0, 0], s_max[0, 0])],
[random.uniform(s_min[1, 0], s_max[1, 0])],
])
print("Sampled initial state is:\n {}".format(x0))
Q = np.matrix("1 0 ; 0 1")
R = np.matrix(".0005")
x_min = np.array([[-1.5], [-1.5]])
x_max = np.array([[1.5], [1.5]])
u_min = np.array([[-1.]])
u_max = np.array([[1.]])
env = Environment(A, B, u_min, u_max, s_min, s_max, x_min, x_max, Q, R)
args = {
'actor_lr': 0.000001,
'critic_lr': 0.00001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 5000
}
actor = DDPG(env, args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0] + '/'
linear_func_model_name = 'K.model'
model_path = model_path + linear_func_model_name + '.npy'
shield = Shield(env, actor, model_path, force_learning=retrain_shield)
shield.train_shield(learning_method,
number_of_rollouts,
simulation_steps,
explore_mag=0.0004,
step_size=0.0005)
if shield_test:
shield.test_shield(test_episodes, 5000, mode="single")
def cap_attack(bots, cap):
global money
for b in bots:
if cap.range >= math.sqrt(
(cap.rect.centerx - b.rect.centerx)**2 +
(cap.rect.centery - b.rect.centery)**2
) and cap.rect.x != cap.x and not cap.moving and cap.shield == None:
cap.shield = Shield(cap.rect.x, cap.rect.y)
cap.direction = direction(cap, b)
if cap.shield != None:
money += cap.shield.throw(cap, bots, DISPLAYSURF)
cap.change_image()
def prep_shield_count(self): """Show how many shields are left""" # Position the shield_image below the ship_image self.shield_image = Group() shield = Shield(self.ai_settings, self.screen, size='mini') shield.rect.x = 20 shield.rect.y = 20 + self.ship.height self.shield_image.add(shield) # Display the shield count image self.display_shield_count() #Position the shield count image self.shield_count_rect = self.shield_count.get_rect() self.shield_count_rect.y = shield.rect.y + 10 self.shield_count_rect.left = shield.rect.right + 10
def test_shield():
shield = Shield ("192.168.1.1")
assert shield.getShieldSentence() == False
shield2 = Shield("192.168.1.1")
assert shield2.getShieldSentence() == False
shield3 =Shield("192.168.1.1")
shield4 = Shield("192.168.1.1")
shield5=Shield("192.168.1.1")
shield6=Shield("192.168.1.1")
shield7=Shield("192.168.1.1")
assert shield7.getShieldSentence() == True
def cooling (learning_method, number_of_rollouts, simulation_steps, learning_eposides, critic_structure, actor_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100):
A = np.matrix([
[1.01,0.01,0],
[0.01,1.01,0.01],
[0.0,0.01,1.01]])
B = np.matrix([
[1, 0, 0],
[0, 1, 0],
[0, 0, 1]])
#intial state space
s_min = np.array([[ 1.6],[ 1.6], [1.6]])
s_max = np.array([[ 3.2],[ 3.2], [3.2]])
Q = np.eye(3)
R = np.eye(3)
x_min = np.array([[-3.2],[-3.2],[-3.2]])
x_max = np.array([[3.2],[3.2],[3.2]])
u_min = np.array([[-1.],[-1.],[-1.]])
u_max = np.array([[ 1.],[ 1.],[ 1.]])
env = Environment(A, B, u_min, u_max, s_min, s_max, x_min, x_max, Q, R, bad_reward=-1000)
args = { 'actor_lr': 0.001,
'critic_lr': 0.01,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 5000}
actor = DDPG(env, args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0]+'/'
linear_func_model_name = 'K.model'
model_path = model_path+linear_func_model_name+'.npy'
names = {0:"cart position, meters", 1:"cart velocity", 2:"pendulum angle, radians", 3:"pendulum angle velocity"}
shield = Shield(env, actor, model_path, force_learning=retrain_shield)
shield.train_shield(learning_method, number_of_rollouts, simulation_steps, names=names, explore_mag = 0.02, step_size = 0.0025)
if shield_test:
shield.test_shield(test_episodes, 5000, mode="single")
def readFile(screen, blocks, shield, powerpills, intersections):
file = open("images/otherpacmanportalmaze.txt", "r")
contents = file.read()
line = ''
all_lines = []
for chars in contents:
if chars != '\n':
line += chars
else:
all_lines.append(line)
line = ''
i = 0
j = 0
intersection_num = 0
for rows in all_lines:
for chars in rows:
if chars == 'X':
new = Blocks(screen)
new.rect.x, new.rect.y = 13 * i, 13 * j
blocks.add(new)
elif chars == 'd':
thepowerpill = Powerpills(screen)
thepowerpill.rect.x, thepowerpill.rect.y = 13 * i, 13 * j
powerpills.add(thepowerpill)
elif chars == 'b':
thepowerpill = Powerpills(screen, 'big')
thepowerpill.rect.x, thepowerpill.rect.y = 13 * i, 13 * j
powerpills.add(thepowerpill)
elif chars == 'i':
intersection = Intersections(screen, intersection_num)
intersection_num += 1
intersection.rect.x, intersection.rect.y = 13 * i, 13 * j
intersections.add(intersection)
elif chars == 'o':
theshield = Shield(screen)
theshield.rect.x, theshield.rect.y = 13 * i, 13 * j
shield.add(theshield)
i += 1
i = 0
j += 1
def cruise (learning_method, number_of_rollouts, simulation_steps,learning_eposides, critic_structure, actor_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100):
A = np.matrix([[0.99501]])
B = np.matrix([[0.0078125]])
#intial state space
s_min = np.array([[-1.0]])
s_max = np.array([[ 1.0]])
Q = np.matrix("1")
R = np.matrix(".0005")
x_min = np.array([[-1.5]])
x_max = np.array([[ 1.5]])
u_min = np.array([[-10.]])
u_max = np.array([[ 10.]])
env = Environment(A, B, u_min, u_max, s_min, s_max, x_min, x_max, Q, R)
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 5000}
actor = DDPG(env, args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0]+'/'
linear_func_model_name = 'K.model'
model_path = model_path+linear_func_model_name+'.npy'
shield = Shield(env, actor, model_path, force_learning=retrain_shield, debug=False)
shield.train_shield(learning_method, number_of_rollouts, simulation_steps, eq_err=0, explore_mag = 1.0, step_size = 1.0)
if shield_test:
shield.test_shield(test_episodes, 5000, mode="single")
def readFile(screen, blocks, shield, powerpills, portal):
file = open("images/otherpacmanportalmaze.txt", "r")
contents = file.read()
line = ''
all_lines = []
for chars in contents:
if chars != '\n':
line += chars
else:
all_lines.append(line)
line = ''
i = 0
j = 0
for rows in all_lines:
for chars in rows:
if chars == 'X':
new = Blocks(screen)
new.rect.x, new.rect.y = 13 * i, 13 * j
blocks.add(new)
elif chars == 'd':
thepowerpill = Powerpills(screen)
thepowerpill.rect.x, thepowerpill.rect.y = 13 * i, 13 * j
powerpills.add(thepowerpill)
elif chars == 'o':
theshield = Shield(screen)
theshield.rect.x, theshield.rect.y = 13 * i, 13 * j
shield.add(theshield)
# This is where the horizontal portal is supposed to be
elif chars == 'h':
pass
# Vertical portal?
elif chars == 'P':
theportal = Portal(screen)
theportal.rect.x, theportal.rect.y = 13 * i, 13 * j
portal.add(theportal)
pass
i += 1
i = 0
j += 1
def create_shields(cls, *groups):
"""Create the shields which block bullets.
Arguments:
groups (pygame.sprite.Group): The groups the shields will be in.
Returns:
list [Shield]: The shields that were created.
"""
shields = []
shield_gap = int((DISPLAY.width - NUM_SHIELDS * SHIELD.width) /
(NUM_SHIELDS + 1))
start = shield_gap
end = DISPLAY.width - SHIELD.width
step = SHIELD.width + shield_gap
for pos_x in range(start, end, step):
shields.append(Shield((pos_x, SHIELD_HEIGHT), *groups))
return shields
def readFile(screen, blocks, shield, pellets, powerpellets):
file = open("maze.txt", "r")
contents = file.read()
line = ''
all_lines = []
for chars in contents:
if chars != '\n':
line += chars
else:
all_lines.append(line)
line = ''
i = 0
j = 0
for rows in all_lines:
for chars in rows:
if chars == 'X':
new = Bricks(screen)
new.rect.x, new.rect.y = 13 * i, 13 * j
blocks.add(new)
elif chars == 'P':
thepowerpellets = PowerPellets(screen)
thepowerpellets.rect.x, thepowerpellets.rect.y = 13 * i, 13 * j
powerpellets.add(thepowerpellets)
elif chars == '-':
theshield = Shield(screen)
theshield.rect.x, theshield.rect.y = 13 * i, 13 * j
shield.add(theshield)
elif chars == 'o':
thepellets = Pellets(screen)
thepellets.rect.x, thepellets.rect.y = 13 * i, 13 * j
pellets.add(thepellets)
elif chars == 'T':
pass
i += 1
i = 0
j += 1
def create_shield(ai_settings, screen, stats, shield): """Add one shield to group Shield""" if (stats.alien_kills % ai_settings.shield_drop_point == 0): if (len(shield) < 1): new_shield = Shield(ai_settings, screen) shield.add(new_shield)
def __init__(self):
##THESE LINES JUST HOLD HARDCODED LISTS AND DICTS SO TO FIND OTHER ATTRS SCROLL DOWN FOR THE NEXT UNINDENTED COMMENT
#list of classes with ships in them for the player to play
#use the players faction as a DICT key to view the list of playable ships
self.playerShips = {
"UFP": [ #Could add the warship voyager
Ship("Enterprise A", "UFP", 1000,
Shield("Standard Issue Shield", 150, 12), [
Weapon("Standard Phasers", 100, "Energy"),
Weapon("Photon Torpedo", 150, "Antimatter")
], 100),
Ship("Enterprise B", "UFP", 1400,
Shield("Standard Issue Shield", 300, 24), [
Weapon("Standard Phasers", 140, "Energy"),
Weapon("Photon Torpedo", 190, "Antimatter")
], 125),
Ship("Enterprise C", "UFP", 1750,
Shield("Standard Issue Shield", 450, 36), [
Weapon("Standard Phasers", 180, "Energy"),
Weapon("Photon Torpedo", 230, "Antimatter")
], 200),
Ship("Enterprise D", "UFP", 2000,
Shield("Standard Issue Shield", 600, 48), [
Weapon("Standard Phasers", 215, "Energy"),
Weapon("Photon Torpedo", 255, "Antimatter")
], 220),
Ship("Enterprise E", "UFP", 3200,
Shield("Standard Issue Shield", 900, 60), [
Weapon("Standard Phasers", 280, "Energy"),
Weapon("Photon Torpedo", 320, "Antimatter"),
Weapon("Quantum Torpedo", 570, "Special")
], 260),
Ship("Voyager", "UFP", 2200,
Shield("Standard Issues Shield", 650, 32), [
Weapon("Standard Phasers", 245, "Energy"),
Weapon("Photon Torpedo", 260, "Antimatter")
], 200),
Ship("Defiant", "UFP", 3000,
Shield("Standard Issues Shield", 900, 60), [
Weapon("Pulse Phasers", 250, "Energy"),
Weapon("Photon Torpedo", 320, "Antimatter"),
Weapon("Quantum Torpedo", 570, "Special")
], 200),
Ship("Polaris", "UFP", 4000,
Shield("Standard Issues Shield", 1100, 60), [
Weapon("Standard Phasers", 280, "Energy"),
Weapon("Photon Torpedo", 350, "Antimatter"),
Weapon("Quantum Torpedo", 630, "Special")
], 250)
],
"Cardassian": [
Ship("Galor Class", "Cardassian", 1800,
Shield("Standard Shield", 550, 22), [
Weapon("Phaser Array", 200, "Energy"),
Weapon("Antimatter Torpedo", 245, "Antimatter")
], 200),
Ship("Hideki Class", "Cardassian", 1400,
Shield("Standard Shield", 250, 15), [
Weapon("Phaser Array", 150, "Energy"),
Weapon("Antimatter Torpedo", 230, "Antimatter")
], 140),
Ship("Kutet Class", "Cardassian", 2200,
Shield("Standard Shield", 900, 30), [
Weapon("Phasr Array", 340, "Energy"),
Weapon("Antimatter Torpedo", 270, "Antimatter")
], 250),
],
"Klingon": [
Ship("B'rel", "Klingon", 1200,
Shield("Standard Shield", 250, 26), [
Weapon("Disruptors", 135, "Energy"),
Weapon("Photon Torpedo", 245, "Antimatter")
], 100),
Ship("Vor'cha", "Klingon", 1600,
Shield("Standard Shield", 400, 26), [
Weapon("Disruptors", 180, "Energy"),
Weapon("Photon Torpedo", 300, "Antimatter")
], 150),
Ship("K'vort", "Klingon", 2000,
Shield("Standard Shield", 550, 32), [
Weapon("Disruptors", 215, "Energy"),
Weapon("Photon Torpedo", 330, "Antimatter")
], 175),
Ship("Negh'Var", "Klingon", 2400,
Shield("Standard Shield", 700, 38), [
Weapon("Disruptors", 260, "Energy"),
Weapon("Photon Torpedo", 360, "Antimatter")
], 200),
Ship("Sword of Kahless", "Klingon", 3000,
Shield("Standard Shield", 900, 50), [
Weapon("Disruptors", 300, "Energy"),
Weapon("Photon Torpedo", 400, "Antimatter")
], 250)
],
"Romulan": [
Ship("D7 Class", "Romulan", 1250,
Shield("Standard Shield", 300, 20), [
Weapon("Tri-Phasic Emitters", 180, "Energy"),
Weapon("Plasma Torpedo", 245, "Plasma")
], 110),
Ship("Valdore", "Romulan", 2000,
Shield("Standard Shield", 500, 38), [
Weapon("Tri-Phasic Emitters", 225, "Energy"),
Weapon("Plasma Torpedo", 300, "Plasma")
], 250),
Ship("D'deridex", "Romulan", 2500,
Shield("Standard Shield", 800, 56), [
Weapon("Tri-Phasic Emitters", 270, "Energy"),
Weapon("Plasma Torpedo", 360, "Plasma")
], 250),
Ship("Scimitar", "Romulan", 3600,
Shield("Standard Shield", 1100, 73), [
Weapon("Tri-Phasic Emitters", 330, "Energy"),
Weapon("Plasma Torpedo", 400, "Plasma"),
Weapon("Thalaron Pulse", 700, "Special")
], 250)
],
"Dominion": [
Ship("Jem'Hadar Cruiser", "Dominion", 1250,
Shield("Polaron Shield", 300, 20), [
Weapon("Phased Polaron Beam", 180, "Polaron"),
Weapon("Polaron Torpedo", 245, "Polaron")
], 110),
Ship("Chimera Battlecruiser", "Dominion", 2200,
Shield("Polaron Shield", 700, 30), [
Weapon("Phased Polaron Beam", 250, "Polaron"),
Weapon("Polaron Torpedo", 260, "Polaron")
], 110),
Ship("Hydra Dreadnought", "Dominion", 3000,
Shield("Polaron Shield", 700, 40), [
Weapon("Phased Polaron Beam", 320, "Polaron"),
Weapon("Polaron Torpedo", 280, "Polaron")
], 110)
],
"Borg": [
Ship("Borg Sphere", "Borg", 2500, Shield("No Shield", 0, 0), [
Weapon("Borg Cutting Beam", 300, "Energy"),
Weapon("Shield Drain Torpedo", 400, "Drain")
], 100),
Ship("Borg Cube", "Borg", 5000, Shield("No Shield", 0, 0), [
Weapon("Borg Cutting Beam", 700, "Energy"),
Weapon("Shield Drain Torpedo", 800, "Drain")
], 150),
Ship("Borg Diamond", "Borg", 3300, Shield("No Shield", 0, 0), [
Weapon("Borg Cutting Beam", 500, "Energy"),
Weapon("Shield Drain Torpedo", 600, "Drain")
], 125),
Ship("Progenitor", "Borg", 92000, Shield("No Shield", 0, 0), [
Weapon("Borg Cutting Beam", 20000, "Energy"),
Weapon("Ragnarok Torpedo", 50000, "Dark Matter"),
Weapon("Reaper Beam", 90000, "Polaron")
], 500)
] #Reaper beam is not actually Polaron based, just needs it here to bypass shields
}
#The Keys in this list also double up as the playable factions
#as well as the keys in playerShips
self.alliances = {
"UFP": "UFP;Bajoran;",
"Klingon": "Klingon;",
"Cardassian": "Cardassian;Dominion;",
"Dominion": "Dominion;Cardassian;Breen;",
"Romulan": "",
"Borg": ""
}
#These will be possible factions for other ships
self.factions = [
"UFP", "Klingon", "Bajoran", "Cardassian", "Dominion", "Breen",
"Hirogen", "Borg", "Romulan", "Orion Pirates", "Nausicaans",
"Aaamazzarite", "Acamarian", "Aenar", "Akritirian", "Allasomorph",
"Antaran", "Berellian", "Ferengi", "Horta", "Kazon", "Ocampa",
"Species 8472", "Tholian", "Tellarite", "Viidians"
]
## THIS IS WHERE OTHER ATTRS WILL BE STORED SUCH AS THE PLAYER AND MAP OBJECTS
self.player = self.setupPlayer() # sets up the player object
self.setupGame() # sets up the game difficulty and creates the gameMap
self.lastAction = None # a variable to hold the last action the player performed
self.activeShips = {
} #holds dict of active Enemy objects as {name STR:OBJ(Enemy)}
self.enemyThreads = []
def car(learning_method,
number_of_rollouts,
simulation_steps,
learning_episodes,
actor_structure,
critic_structure,
train_dir,
nn_test=False,
retrain_shield=False,
shield_test=False,
test_episodes=100,
retrain_nn=False,
safe_training=False,
shields=1,
episode_len=100,
penalty_ratio=0.1):
# Set up the actual model
def f(x, u):
return np.matrix([[x[1, 0]],
[0.001 * u[0, 0] - 0.0025 * np.cos(3 * x[0, 0])]])
def f_to_str(K):
kstr = K_to_str(K)
f = []
f.append("x[2]")
f.append("0.001*{} - 0.0025 * cos(3 * x[1])".format(kstr[0]))
return f
def rewardf(x, Q, u, R):
return x[0, 0] - 0.6
def testf(x, u):
return x[0, 0] < -np.pi / 3
def terminalf(x):
return x[0, 0] >= 0.6
x_min = np.array([[-1.2], [-0.007]])
x_max = np.array([[0.7], [0.007]])
s_min = np.array([[-0.5], [0.0]])
s_max = np.array([[-0.5], [0.0]])
u_min = np.array([[-1.0]])
u_max = np.array([[1.0]])
# Set up a linearized model
# We'll use splits at -0.1 to 0.1 around each peak (since the dynamics
# use cos(3x), the peaks are at multiples of pi / 3).
breaks = [-np.pi / 3 - 0.2, -np.pi / 3 + 0.2, -0.2, 0.2, np.pi / 3 - 0.2]
break_breaks = [5, 5, 5]
mins = [
-np.cos(-np.pi - 0.6), -np.cos(-np.pi + 0.6), -1, -1,
np.cos(np.pi - 0.6)
]
maxes = [1, 1, -np.cos(-0.6), -np.cos(0.6), 1]
lower_As = []
upper_As = []
B = np.array([[0.0], [0.001]])
for i in range(len(breaks) - 1):
max_m = (maxes[i + 1] - maxes[i]) / (breaks[i + 1] - breaks[i])
min_m = (mins[i + 1] - mins[i]) / (breaks[i + 1] - breaks[i])
# lA * x + B * u <= x' <= uA * x + B * u
lower_As.append(np.matrix([[0.0, 1.0], [0.0025 * min_m, 0.0]]))
upper_As.append(np.matrix([[0.0, 1.0], [0.0025 * max_m, 0.0]]))
Bs = [B] * len(lower_As)
# We consider unsafe behavior to be moving over the left side of the hill
uA = np.matrix([[1.0, 0.0]])
ub = np.matrix([[-np.pi / 3]])
env = PolySysEnvironment(f,
f_to_str,
rewardf,
testf,
None,
2,
1,
None,
None,
s_min,
s_max,
x_min=x_min,
x_max=x_max,
u_min=u_min,
u_max=u_max,
timestep=1.0,
terminalf=terminalf,
unsafe_A=uA,
unsafe_b=ub,
approx=True,
breaks=breaks,
break_breaks=break_breaks,
lower_As=lower_As,
lower_Bs=Bs,
upper_As=upper_As,
upper_Bs=Bs)
if retrain_nn:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': episode_len,
'max_episodes': learning_episodes,
'minibatch_size': 64,
'random_seed': 6554, # 6553
'tau': 0.005,
'model_path': train_dir + "retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 500
}
else:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': episode_len,
'max_episodes': 0,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 500
}
Ks = [np.matrix([[0.0, 0.0]])]
invs = [(np.matrix([[1, 0], [-1, 0], [0, 1],
[0, -1]]), np.matrix([[0.8], [0.8], [0.07], [0.07]]))]
covers = [(invs[0][0], invs[0][1], np.matrix([[-0.8], [-0.07]]),
np.matrix([[0.8], [0.07]]))]
bound = 30
initial_shield = Shield(env,
K_list=Ks,
inv_list=invs,
cover_list=covers,
bound=bound)
actor, shield = DDPG(env,
args,
safe_training=safe_training,
shields=shields,
initial_shield=initial_shield,
penalty_ratio=penalty_ratio)
if shield_test:
shield.test_shield(actor, test_episodes, 5000)
actor.sess.close()
def obstacle(learning_method, number_of_rollouts, simulation_steps,
learning_episodes, actor_structure, critic_structure, train_dir,
nn_test=False, retrain_shield=False, shield_test=False,
test_episodes=100, retrain_nn=False, safe_training=False, shields=1,
episode_len=100, penalty_ratio=0.1):
A = 0.2 * np.matrix([[0.0, 0.0, 10.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 10.0, 0.0],
[0.0, 0.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 0.0, 0.0]])
B = 0.5 * np.matrix([[0.0, 0.0], [0.0, 0.0], [10.0, 0.0], [0.0, 10.0], [0.0, 0.0]])
#s_min = np.array([[-0.1], [-0.1], [-0.1], [-0.1], [1]])
#s_max = np.array([[0.1], [0.1], [0.1], [0.1], [1]])
s_min = np.array([[0.0], [0.0], [0.0], [0.0], [1.0]])
s_max = np.array([[0.0], [0.0], [0.0], [0.0], [1.0]])
x_goal = 3.0
y_goal = 0.0
x_min = np.array([[-10.0], [-10.0], [-20.0], [-20.0], [0.0]])
x_max = np.array([[ 10.0], [ 10.0], [ 20.0], [ 20.0], [2.0]])
def rewardf(x, u):
return -(abs(x[0,0] - x_goal) + abs(x[1,0] - y_goal))
def terminalf(x):
return x[0,0] >= x_goal and x[1,0] >= y_goal
u_min = np.array([[-2.0], [-2.0]])
u_max = np.array([[ 5.0], [ 5.0]])
# There is a box at [0, 2] to [1, 3] which is unsafe
unsafe_A = [np.matrix([[ 1.0, 0.0, 0.0, 0.0, 0.0],
[-1.0, 0.0, 0.0, 0.0, 0.0],
[ 0.0, 1.0, 0.0, 0.0, 0.0],
[ 0.0, -1.0, 0.0, 0.0, 0.0]])]
unsafe_b = [np.matrix([[1.0], [0.0], [3.0], [-2.0]])]
env = Environment(A, B, u_min, u_max, s_min, s_max, x_min, x_max,
None, None, continuous=True, rewardf=rewardf, terminalf=terminalf,
unsafe_A=unsafe_A, unsafe_b=unsafe_b)
if retrain_nn:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': episode_len,
'max_episodes': learning_episodes, # originally 1000
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 5000}
else:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': episode_len,
'max_episodes': learning_episodes,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 5000}
Ks = [np.matrix([[0.0, 0.0, 0.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0, 0.0]]),
np.matrix([[0.0, 0.0, 0.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0, 0.0]])]
invs = [(np.matrix([[-1.0, 0.0, 0.0, 0.0, 0.0],
[ 0.0, 0.0, -1.0, 0.0, 0.0]]),
np.matrix([[-1.5], [1.25]])),
(np.matrix([[0.0, 1.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 1.0, 0.0]]),
np.matrix([[1.5], [1.25]]))]
covers = [(invs[0][0], invs[0][1],
np.matrix([[1.5, -0.5, -0.3, -0.3, 1.0]]),
np.matrix([[x_goal + 0.5, y_goal + 0.5, 2.0, 2.0, 1.0]])),
(invs[1][0], invs[1][1],
np.matrix([[-0.5, -0.5, -0.3, -0.3, 1.0]]),
np.matrix([[x_goal + 0.1, 0.5, 2.0, 1.0, 1.0]]))]
bound = 8
initial_shield = Shield(env, K_list=Ks, inv_list=invs, cover_list=covers,
bound=bound)
actor, shield = DDPG(env, args, safe_training=safe_training, shields=shields,
initial_shield=initial_shield, penalty_ratio=penalty_ratio, bound=bound)
if shield_test:
shield.test_shield(actor, test_episodes, 5000)
actor.sess.close()
alf = Enemy("{} ship".format(factionChoice),factionChoice,hull,s,[w1,w2])
return alf
#----------TESTS----------#
#from enemy import Enemy
#from ship import Ship, voy
##nx = Enemy("Enterprise NX","UFP",1500,voy.shield,voy.weapons)
##
weyWep1 = Weapon("Phased Polaron Beam",213,"Polaron")
weyWep2 = Weapon("Polaron Torpedo",297,"Polaron")
weyShield = Shield("Polaron Shield",1000,12)
weyoun = Enemy("Hydra Battlecruiser","Dominion",4000,weyShield,[weyWep1,weyWep2])
##
##
##def fightNX():
## voy.raiseShields()
## while voy.alive and nx.alive:
## choice = random.randint(0,1)
## nx.receiveDamage(voy.weapons[choice])
## voy.receiveDamage(nx.weapons[choice])
## nx.updateShields()
##
## if voy.alive:
## print("Voy won")
## else:
## print("NX won")
def biology (learning_method, number_of_rollouts, simulation_steps, learning_eposides, critic_structure, actor_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100, retrain_nn=False):
# 10-dimension and 1-input system and 1-disturbance system
ds = 3
us = 2
#Dynamics that are defined as a continuous function!
def f (x, u):
#random disturbance
#d = random.uniform(0, 20)
delta = np.zeros((ds, 1), float)
delta[0,0] = -0.01*x[0,0] - x[1,0]*(x[0,0]+4.5) + u[0,0]
delta[1,0] = -0.025*x[1,0] + 0.000013*x[2,0]
delta[2,0] = -0.093*(x[2,0] + 15) + (1/12)*u[1,0]
return delta
#Closed loop system dynamics to text
def f_to_str(K):
kstr = K_to_str(K)
f = []
f.append("-0.01*x[1] - x[2]*(x[1]+4.5) + {}".format(kstr[0]))
f.append("-0.025*x[2] + 0.000013*x[3]")
f.append("-0.093*(x[3] + 15) + (1/12)*{}".format(kstr[1]))
return f
h = 0.01
# amount of Gaussian noise in dynamics
eq_err = 1e-2
#intial state space
s_min = np.array([[-2],[-0],[-0.1]])
s_max = np.array([[ 2],[ 0],[ 0.1]])
Q = np.zeros((ds,ds), float)
R = np.zeros((us,us), float)
np.fill_diagonal(Q, 1)
np.fill_diagonal(R, 1)
#user defined unsafety condition
def unsafe_eval(x):
if (x[0,0] >= 5):
return True
return False
def unsafe_string():
return ["x[1] - 5"]
def rewardf(x, Q, u, R):
reward = 0
reward += -np.dot(x.T,Q.dot(x))-np.dot(u.T,R.dot(u))
if (unsafe_eval(x)):
reward -= 100
return reward
def testf(x, u):
if (unsafe_eval(x)):
print x
return -1
return 0
u_min = np.array([[-50.], [-50]])
u_max = np.array([[ 50.], [ 50]])
env = PolySysEnvironment(f, f_to_str,rewardf, testf, unsafe_string, ds, us, Q, R, s_min, s_max, u_max=u_max, u_min=u_min, timestep=h)
############ Train and Test NN model ############
if retrain_nn:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': 1000,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000}
else:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir+"model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000}
actor = DDPG(env, args=args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0]+'/'
linear_func_model_name = 'K.model'
model_path = model_path+linear_func_model_name+'.npy'
shield = Shield(env, actor, model_path=model_path, force_learning=retrain_shield)
shield.train_polysys_shield(learning_method, number_of_rollouts, simulation_steps, eq_err=eq_err, explore_mag = 0.4, step_size = 0.5, aggressive=True, without_nn_guide=True, enable_jit=True)
if shield_test:
shield.test_shield(test_episodes, 1000, mode="single")
actor.sess.close()
class Game(object):
def __init__(self):
## sets dimension of the screen and the background
pygame.init()
self.w = 900
self.h = 600
self.screen = pygame.display.set_mode((self.w, self.h), pygame.FULLSCREEN)
self.background = pygame.image.load("background.png")
## creates all the objects and lists needed for the game
self.player = Player(self.w, self.h, self.background)
self.shooter = Shooter(self.player.x, self.player.y, self.player.d)
self.shield = Shield(self.player.x, self.player.y, self.player.d)
self.generator = Generator(self.player.x, self.player.y, self.player.d)
self.bullet = Bullet(self.shooter.x, self.shooter.y, self.shooter.image_rotation)
self.enemies = [Enemy(self.w, self.h)]
self.enemies_2 = []
self.counter_enemies_2_dead = 0
self.collectible = Collectible(self.w, self.h)
self.ray = Ray(self.w, self.h)
## loads energy image and sets its default value
self.energy_5 = pygame.image.load("energy_5.png")
self.energy = 100
## sets all default values
self.points = 0
self.killed_big_enemies = 0
self.killed_small_enemies = 0
self.rays_collected = 0
self.collectibles_collected = 0
## initializes fonts and creates the first text
font = pygame.font.SysFont("comic", 64)
self.font = pygame.font.SysFont("arial", 10)
text = font.render("Click to play", True, (255, 255, 255))
## loads sound
self.sound = pygame.mixer.Sound("blip.wav")
## sets timer and default timer variables
self.clock = pygame.time.Clock()
self.timer = 0
self.seconds = 0
self.minutes = 0
## manages beginning screen and first inputs
beginning = True
while beginning:
self.screen.fill((0, 0, 0))
self.screen.blit(text, ((self.w / 2) - 150, (self.h / 2) - 64))
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.display.quit()
if event.type == pygame.MOUSEBUTTONUP:
beginning = False
self.screen.blit(self.background, (0, 0))
self.screen.blit(self.player.images[1], (self.player.x, self.player.y))
self.screen.blit(self.shooter.image, (self.shooter.x, self.shooter.y))
self.running()
pygame.display.update()
def running(self):
while True:
## manages time
time_passed = self.clock.tick(30)
time_seconds = time_passed / 1000.0
self.timer += 1
self.seconds += 0.03
if self.seconds >= 60:
self.minutes += 1
self.seconds = 0
## gets all the inputs and calls function related to each input
for event in pygame.event.get():
if event.type == pygame.KEYUP:
if event.key == pygame.K_ESCAPE:
pygame.display.quit()
if event.key == pygame.K_a:
self.shield.index -= 1
if self.shield.index == -5:
self.shield.index = -1
elif event.key == pygame.K_j:
self.shield.index += 1
if self.shield.index == 4:
self.shield.index = 0
elif event.key == pygame.K_l:
self.generator.index += 1
if self.generator.index == 4:
self.generator.index = 0
elif event.key == pygame.K_s:
self.generator.index -= 1
if self.generator.index == -5:
self.generator.index = -1
key = pygame.key.get_pressed()
if key[pygame.K_UP]:
self.player.moves_up(time_seconds)
self.player.index = 0
if key[pygame.K_DOWN]:
self.player.moves_down(time_seconds)
self.player.index = 1
if key[pygame.K_LEFT]:
self.player.moves_left(time_seconds)
self.player.index = 2
if key[pygame.K_RIGHT]:
self.player.moves_right(time_seconds)
self.player.index = 3
## blits the background
self.screen.blit(self.background, (0, 0))
## manages energy and prints it
if self.energy >= 100:
self.energy = 100
if self.energy <= 10:
self.sound.play()
if self.energy <= 0:
self.collectible.finished = True
for n in xrange(0, self.energy / 5):
self.screen.blit(self.energy_5, (self.w - 35 - 10 * n, 4))
## manages the text (points and time)
text = self.font.render((str(self.minutes) + ":" + str(self.seconds)), True, (255, 255, 255))
self.screen.blit(text, (10, 10))
text = self.font.render(("Points: " + str(self.points)), True, (255, 255, 255))
self.screen.blit(text, (440, 10))
## manages collectibles
self.collectible.blit_check(self.background, self.timer)
if self.collectible.blit:
self.screen.blit(self.collectible.image, (self.collectible.x, self.collectible.y))
self.collectible_rect = pygame.Rect(self.collectible.x, self.collectible.y, 20, 20)
## manages player and collision with collectible
self.screen.blit(self.player.images[self.player.index], (self.player.x, self.player.y))
self.player_rect = pygame.Rect(self.player.x, self.player.y, 40, 40)
if self.collectible.blit:
if self.player_rect.colliderect(self.collectible_rect):
i = self.collectible.index
self.collectible = Collectible(self.w, self.h)
self.collectible.index = i
self.points += 100
self.collectibles_collected += 1
## manages bullet, checks hits with walls
self.bullet.moves(self.shooter.x, self.shooter.y)
if self.bullet.rotate:
self.bullet.rotates(time_seconds)
self.screen.blit(self.bullet.rotated_image, self.bullet.image_draw_pos)
self.bullet.check_shot()
if self.bullet.shot:
self.energy -= 5
if self.bullet.shot:
self.screen.blit(self.bullet.rotated_image, (self.bullet.x, self.bullet.y))
self.bullet.hits_borders(self.background)
if self.bullet.hits_bg:
self.bullet = Bullet(self.shooter.x, self.shooter.y, self.shooter.image_rotation)
self.bullet_rect = pygame.Rect(self.bullet.x, self.bullet.y, 4, 10)
## manages generator
self.generator.moves(self.player.x, self.player.y, self.player.d)
self.screen.blit(self.generator.images[self.generator.index], (self.generator.x, self.generator.y))
generator_rect = pygame.Rect(self.generator.x, self.generator.y, self.generator.w, self.generator.h)
## manages shooter
self.shooter.moves(self.player.x, self.player.y, self.player.d)
self.shooter.rotates(time_seconds)
self.screen.blit(self.shooter.rotated_image, self.shooter.image_draw_pos)
## manages shield
self.shield.moves(self.player.x, self.player.y, self.player.d)
self.screen.blit(self.shield.images[self.shield.index], (self.shield.x, self.shield.y))
shield_rect = pygame.Rect(self.shield.x, self.shield.y, self.shield.w, self.shield.h)
## manages big enemies one by one, checks collisions with bullets, shield and tank
for n in xrange(0, len(self.enemies)):
enemy = self.enemies[n]
enemy.moves(time_seconds)
enemy.bounces(self.background)
if enemy.error:
self.enemies[n] = Enemy(self.w, self.h)
self.screen.blit(enemy.image, (enemy.x, enemy.y))
enemy_rect = pygame.Rect(enemy.x, enemy.y, enemy.d, enemy.d)
if enemy_rect.colliderect(self.bullet_rect):
self.bullet = Bullet(self.shooter.x, self.shooter.y, self.shooter.image_rotation)
self.enemies_2.append(Enemy_2(enemy.x, enemy.y))
self.enemies_2.append(Enemy_2(enemy.x, enemy.y))
self.enemies_2.append(Enemy_2(enemy.x, enemy.y))
enemy.alive = False
self.points += 10
self.killed_big_enemies += 1
elif enemy_rect.colliderect(shield_rect):
enemy.reverse()
self.energy -= 5
elif enemy_rect.colliderect(self.player_rect):
enemy.reverse()
self.energy -= 10
if (self.timer == 30 * 5 or self.timer == 30 * 10) and len(self.enemies) <= 2:
self.enemies.append(Enemy(self.w, self.h))
## temporary list to manage the elimination of some enemies
l = []
for n in xrange(0, len(self.enemies)):
enemy = self.enemies[n]
if enemy.alive:
l.append(self.enemies[n])
self.enemies = l
## manages small enemies one by one, checks collision with bullets, shield and tank
for n in xrange(0, len(self.enemies_2)):
enemy_2 = self.enemies_2[n]
enemy_2.moves(time_seconds)
enemy_2.bounces(self.background)
if enemy.error:
self.enemies_2[n] = Enemy_2(self.w, self.h)
self.screen.blit(enemy_2.image, (enemy_2.x, enemy_2.y))
enemy_2_rect = pygame.Rect(enemy_2.x, enemy_2.y, enemy_2.d, enemy_2.d)
if enemy_2_rect.colliderect(self.player_rect):
enemy_2.reverse()
self.energy -= 10
elif enemy_2_rect.colliderect(shield_rect):
enemy_2.reverse()
self.energy -= 5
elif enemy_2_rect.colliderect(self.bullet_rect):
self.bullet = Bullet(self.shooter.x, self.shooter.y, self.shooter.image_rotation)
enemy_2.alive = False
self.counter_enemies_2_dead += 1
self.points += 10
self.killed_small_enemies += 1
## temporary list to manage the elimination of some enemies
l = []
for n in xrange(0, len(self.enemies_2)):
enemy_2 = self.enemies_2[n]
if enemy_2.alive:
l.append(self.enemies_2[n])
self.enemies_2 = l
if self.counter_enemies_2_dead == 3:
self.counter_enemies_2_dead = 0
self.enemies.append(Enemy(self.w, self.h))
## manages rays of energy and collision with generator and tank and manages life time
self.ray.moves(time_seconds)
self.ray.bounces(self.background)
self.screen.blit(self.ray.image, (self.ray.x, self.ray.y))
ray_rect = pygame.Rect(self.ray.x, self.ray.y, self.ray.d, self.ray.d)
if ray_rect.colliderect(generator_rect):
self.ray.check_caught()
if self.ray.caught:
self.ray = Ray(self.w, self.h)
self.energy += 20
self.points += 10
self.rays_collected += 1
else:
self.ray.reverse()
if ray_rect.colliderect(self.player_rect):
self.ray.reverse()
if (
self.timer >= 30 * 15
and self.timer <= 30 * 15.01
or self.timer >= 30 * 30
and self.timer <= 30 * 30.01
or self.timer >= 30 * 45
and self.timer <= 30 * 45.01
or self.timer >= 30 * 60
and self.timer <= 30 * 60.01
or self.timer >= 30 * 75
and self.timer <= 30 * 75.01
or self.timer >= 30 * 90
and self.timer <= 30 * 90.01
or self.timer >= 30 * 105
and self.timer <= 30 * 105.01
or self.timer >= 30 * 120
and self.timer <= 30 * 120.01
):
self.ray = Ray(self.w, self.h)
## manages the end of the loop
if self.collectible.finished == True:
pygame.display.quit()
self.end()
pygame.display.update()
def end(self):
## creates a whole new screen, calculates points and blits results
pygame.init()
screen = pygame.display.set_mode((600, 300), 0, 32)
screen.fill((0, 0, 0))
points = self.seconds + (self.minutes * 60) + self.energy + self.points
if (
self.killed_big_enemies >= 3
and self.killed_small_enemies >= 3
and self.rays_collected >= 3
and self.collectibles_collected >= 4
):
msg = "You did great!!!"
else:
msg = "You could have done better! ;)"
end = True
while end:
font = pygame.font.SysFont("comic", 40)
text = font.render(("Total points: " + str(points)), True, (255, 255, 255))
text1 = font.render(msg, True, (255, 255, 255))
screen.blit(text, (100, 100))
screen.blit(text1, (100, 200))
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.display.quit()
end = False
pygame.display.update()
def lanekeep (learning_method, number_of_rollouts, simulation_steps, learning_eposides, actor_structure, critic_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100, retrain_nn=False):
v0 = 27.7
cf = 133000
cr = 98800
M = 1650
b = 1.59
a = 1.11
Iz = 2315.3
ds = 4
us = 2
disturbance_x_min = np.array([[0],[0],[-0.035],[0]])
disturbance_x_max = np.array([[0],[0],[ 0.035],[0]])
#Dynamics that are defined as a continuous function!
def f (x, u):
rd = random.uniform(-0.6, 0.6)
delta = np.zeros((ds, 1), float)
delta[0,0] = 1*x[1,0] + v0*x[2,0] + random.uniform(disturbance_x_min[0], disturbance_x_max[0]) #lateral displacement
delta[1,0] = (-1*(cf+cr)/(M*v0))*x[1,0] + ((b*cr-a*cf)/(M*v0)-v0)*x[3,0] + (cf/M)*u[0,0] + random.uniform(disturbance_x_min[1], disturbance_x_max[1]) #lateral velocity
delta[2,0] = x[3,0] + random.uniform(disturbance_x_min[2], disturbance_x_max[2]) #error yaw angle
delta[3,0] = ((b*cr-a*cf)/(Iz*v0))*x[1,0] + (-1*(a*a*cf + b*b*cr)/(Iz*v0))*x[3,0] + (a*cf/Iz)*u[1,0] + random.uniform(disturbance_x_min[3], disturbance_x_max[3]) #yaw rate
return delta
#Closed loop system dynamics to text
def f_to_str(K):
kstr = K_to_str(K)
f = []
f.append("1*x[2] + 27.7*x[3] + d[1]")
f.append("(-1*(133000+98800)/(1650*27.7))*x[2] + ((1.59*98800-1.11*133000)/(1650*27.7)-27.7)*x[4] + (133000/1650)*{} + d[2]".format(kstr[0]))
f.append("x[4] + d[3]")
f.append("((1.59*98800-1.11*133000)/(2315.3*27.7))*x[2] + (-1*(1.11*1.11*133000 + 1.59*1.59*98800)/(2315.3*27.7))*x[4] + (1.11*133000/2315.3)*{} + d[4]".format(kstr[1]))
return f
h = 0.01
# amount of Gaussian noise in dynamics
eq_err = 1e-2
#intial state space
s_min = np.array([[ -0.1],[ -0.1], [-0.1], [ -0.1]])
s_max = np.array([[ 0.1],[ 0.1], [ 0.1], [ 0.1]])
Q = np.matrix("1 0 0 0; 0 1 0 0 ; 0 0 1 0; 0 0 0 1")
R = np.matrix(".0005 0; 0 .0005")
#user defined unsafety condition
def unsafe_eval(x):
if (x[0,0] > 0.9 or x[0, 0] < -0.9): # keep a safe distance from the car in front of you
return True
return False
def unsafe_string():
return ["-(x[1]- -0.9)*(0.9-x[1])"]
def rewardf(x, Q, u, R):
reward = 0
reward += -np.dot(x.T,Q.dot(x))-np.dot(u.T,R.dot(u))
if (unsafe_eval(x)):
reward -= 1e-3
return reward
def testf(x, u):
if (unsafe_eval(x)):
return -1
return 0
# Use sheild to directly learn a linear controller
u_min = np.array([[-1]])
u_max = np.array([[1]])
env = PolySysEnvironment(f, f_to_str,rewardf, testf, unsafe_string, ds, us, Q, R, s_min, s_max, u_max=u_max, u_min = u_min, disturbance_x_min=disturbance_x_min, disturbance_x_max=disturbance_x_max, timestep=h)
if retrain_nn:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 1000,
'max_episodes': 1000,
'minibatch_size': 64,
'random_seed': 2903,
'tau': 0.005,
'model_path': train_dir+"retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000}
else:
args = { 'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 1000,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 2903,
'tau': 0.005,
'model_path': train_dir+"model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000}
actor = DDPG(env, args)
model_path = os.path.split(args['model_path'])[0]+'/'
linear_func_model_name = 'K.model'
model_path = model_path+linear_func_model_name+'.npy'
shield = Shield(env, actor, model_path=model_path, force_learning=retrain_shield)
shield.train_polysys_shield(learning_method, number_of_rollouts, simulation_steps, eq_err=eq_err, explore_mag=0.4, step_size=0.5, without_nn_guide=True, aggressive=True)
if shield_test:
shield.test_shield(test_episodes, 1000, mode="single")
from shield import Shield
from config import Time
from bullet import Mandabullet, Dragonbullet
import time
import math
board = Board(40, 500)
scene = Scene(board)
scene.make(board)
manda = Manda(3, 35, board, scene)
dragon = Dragon(485, 33, board, scene)
manda.place_on_map()
stime = time.time()
mbullet = Mandabullet(board, scene, dragon)
dbullet = Dragonbullet(board, scene, manda)
timec = Time((board.get_width() - 170) / 2, 2, 0.5)
shield = Shield()
etime = 0
def gameover(message):
os.system('clear')
print("Score:" + str(manda.get_coins()))
print(message)
quit()
def movemanda():
def alarmhandler(signum, frame):
raise AlarmException
def user_input(timeout=0.05):
def pendulum(learning_eposides, critic_structure, actor_structure, train_dir, learning_method, number_of_rollouts, simulation_steps,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100, retrain_nn=False):
m = 1.17
l = 1.
g = 10.
#Dynamics that are continuous
A = np.matrix([[0., 1.], [g / l, 0.]])
B = np.matrix([[0.], [1. / (m * l**2.)]])
#intial state space
s_min = np.array([[-0.35], [-0.35]])
s_max = np.array([[0.35], [0.35]])
#reward function
Q = np.matrix([[1., 0.], [0., 1.]])
R = np.matrix([[.005]])
#safety constraint
x_min = np.array([[-0.5], [-0.5]])
x_max = np.array([[0.5], [0.5]])
u_min = np.array([[-15.]])
u_max = np.array([[15.]])
env = Environment(A,
B,
u_min,
u_max,
s_min,
s_max,
x_min,
x_max,
Q,
R,
continuous=True)
if retrain_nn:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 500,
'max_episodes': 1000,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 3800
}
else:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 500,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 3800
}
actor = DDPG(env, args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0] + '/'
linear_func_model_name = 'K.model'
model_path = model_path + linear_func_model_name + '.npy'
def rewardf(x, Q, u, R):
return np.matrix([[env.reward(x, u)]])
shield = Shield(env,
actor,
model_path,
force_learning=retrain_shield,
debug=False)
shield.train_shield(learning_method,
number_of_rollouts,
simulation_steps,
rewardf=rewardf,
eq_err=1e-2,
explore_mag=0.3,
step_size=0.3)
if shield_test:
shield.test_shield(test_episodes, 3800, mode="single")
actor.sess.close()
def magneticpointer (learning_method, number_of_rollouts, simulation_steps,learning_eposides, critic_structure, actor_structure, train_dir,\
nn_test=False, retrain_shield=False, shield_test=False, test_episodes=100, retrain_nn=False):
A = np.matrix([[2.6629, -1.1644, 0.66598], [2, 0, 0], [0, 0.5, 0]])
B = np.matrix([[0.25], [0], [0]])
#intial state space
s_min = np.array([[-1.0], [-1.0], [-1.0]])
s_max = np.array([[1.0], [1.0], [1.0]])
Q = np.matrix("1 0 0 ; 0 1 0; 0 0 1")
R = np.matrix("1")
x_min = np.array([[-3.5], [-3.5], [-3.5]])
x_max = np.array([[3.5], [3.5], [3.5]])
u_min = np.array([[-15.]])
u_max = np.array([[15.]])
env = Environment(A, B, u_min, u_max, s_min, s_max, x_min, x_max, Q, R)
if retrain_nn:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': 1000,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "retrained_model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000
}
else:
args = {
'actor_lr': 0.0001,
'critic_lr': 0.001,
'actor_structure': actor_structure,
'critic_structure': critic_structure,
'buffer_size': 1000000,
'gamma': 0.99,
'max_episode_len': 100,
'max_episodes': learning_eposides,
'minibatch_size': 64,
'random_seed': 6553,
'tau': 0.005,
'model_path': train_dir + "model.chkp",
'enable_test': nn_test,
'test_episodes': test_episodes,
'test_episodes_len': 1000
}
actor = DDPG(env, args)
#################### Shield #################
model_path = os.path.split(args['model_path'])[0] + '/'
linear_func_model_name = 'K.model'
model_path = model_path + linear_func_model_name + '.npy'
def rewardf(x, Q, u, R):
return np.matrix([env.reward(x, u)])
shield = Shield(env,
actor,
model_path,
force_learning=retrain_shield,
debug=False)
shield.train_shield(learning_method,
number_of_rollouts,
simulation_steps,
rewardf=rewardf,
eq_err=0,
explore_mag=0.04,
step_size=0.05,
without_nn_guide=True)
if shield_test:
shield.test_shield(test_episodes, 1000, mode="single")
