def display():
pygame.init()
clock = pygame.time.Clock()
screen = pygame.display.set_mode((SCREEN_WIDTH, SCREEN_HEIGHT), 0, 32)
surface = pygame.Surface(screen.get_size())
surface = surface.convert()
drawGrid(surface)
player = Snake(0)
player.lives = 10
enemy = Snake(1)
enemy.positions = [(random.randint(0,
SIZE - 1), random.randint(0, SIZE - 1))]
food = Food([player])
myfont = pygame.font.SysFont("bahnschrift", 20)
iterations = 0
while (player.lives > 0):
for event in pygame.event.get():
if event.type == pygame.KEYDOWN:
if event.key == pygame.K_1:
print("SPace")
clock.tick(0)
#print(get_food_distance(player, food))
#getObsSmall([player, enemy], food)
clock.tick(1)
drawGrid(surface)
enemy.handle_keys()
obs = (player - food)
# no random action in display phase
action_space = np.array(q_table[obs]).copy()
player.action(action_space, "QL")
enemy.move()
handleSnakeCollisions(player, enemy)
handleFoodEating([player, enemy], food)
getObsGrid([player, enemy], food, size=7, fullGrid=True)
player.draw(surface)
enemy.draw(surface)
food.draw(surface)
screen.blit(surface, (0, 0))
text1 = myfont.render("Score AI {0}".format(player.score), 1,
(250, 250, 250))
text2 = myfont.render("Score Player {0}".format(enemy.score), 1,
(250, 250, 250))
screen.blit(text1, (5, 10))
screen.blit(text2, (SCREEN_WIDTH - 175, 10))
pygame.display.update()
print("Snake Player Final Score:", player.score)
def render(agent, obs_type: str, board_size: int):
pygame.init()
clock = pygame.time.Clock()
screen = pygame.display.set_mode((SCREEN_WIDTH, SCREEN_HEIGHT), 0, 32)
surface = pygame.Surface(screen.get_size())
surface = surface.convert()
drawGrid(surface)
player = Snake(0)
player.lives = 5
enemy = Snake(1)
enemy.lives = 5
enemy.positions = [[5,5]]
food = Food([player])
myfont = pygame.font.SysFont("bahnschrift", 20)
iterations = 0
while (player.lives > 4 and enemy.lives > 4):
clock.tick(4)
drawGrid(surface)
enemyMovement(food=food,enemy=enemy)
# no random action in display phase
if obs_type == "Grid":
obs = getObsGrid([player], food, size=board_size, fullGrid=False)
if obs_type == 'Small':
obs = getObsSmall([player, enemy], food)
action = agent.choose_action(obs)
action_space = [action]
player.action(action_space, "PG")
handleSnakeCollisions(player,enemy)
handleFoodEating([player, enemy], food)
player.draw(surface)
enemy.draw(surface)
food.draw(surface)
screen.blit(surface, (0, 0))
text1 = myfont.render("Score AI {0}".format(player.score), 1, (250, 250, 250))
#text2 = myfont.render("Score Player {0}".format(enemy.score), 1, (250, 250, 250))
screen.blit(text1, (5, 10))
#screen.blit(text2, (SCREEN_WIDTH - 175, 10))
pygame.display.update()
print("Snake Player Final Score:", player.score)
class MyTestCase(unittest.TestCase):
def test_something(self):
self.assertEqual(True, False)
def setUp(self):
print("start test")
self.butter = Food(4, 2, 1)
def tearDown(self):
print("test completed")
def test_food(self):
self.assertEqual(self.butter.energy_food(), 32)
def test_food_type(self):
self.assertFalse(self.butter.energy_food() > 10)
def test_food_equal(self):
self.assertEqual(self.butter.energy_food(), 34.2)
def test_energy(self):
self.assertFalse(self.butter.energy_food() == 0)
obs = getObsSmall(snakes, food)
if player.reward == -DEATH_PENALTY:
done = True
info = ""
return obs, player.reward, done, info
score_history = []
score = 0
n_steps_history = []
for i in tqdm(range(num_episodes)):
player = Snake(0)
enemy = Snake(1)
enemy.positions = [(random.randint(0, SIZE - 1), random.randint(0, SIZE - 1))]
food = Food([player])
done = False
score = 0
# returns a numpy array of the state we care about
observation = getObsGrid(snakes=[player], food=food, size=OBS_GRID_SIZE, fullGrid=False)
#observation = getObsSmall([player, enemy], food)
n_steps = 0
while not done and n_steps < 100:
n_steps += 1
# action needs to be either 0,1,2 or 3
action = agentAC.choose_action(observation)
observation_, reward, done, info = step(snakes=[player, enemy], food=food, action=action, obs_type=obs_type)
agentAC.learn(observation, reward, observation_, done) # For Actor-Critic
#agent.store_rewards(reward) # For REINFORCE
observation = observation_
score += reward
def test_food_type(self):
bread = Food(0, 0, 0)
assert bread.energy_food() == 0
def test_food(self):
butter = Food(4, 2, 2)
assert butter.energy_food() == 43.2
def test_energy(self):
cabbage = Food(22, 92, 1)
assert cabbage.energy_food() == 52
def test_food_double(self):
nuts = Food(44, 93, 23)
food = nuts * 2
assert food.energy_food() != 0
def test_food_equal(self):
apple = Food(34, 2, 11)
assert apple.energy_food() != 3
# for x1 in range(-SIZE+1, SIZE):
# for y1 in range(-SIZE+1, SIZE):
# q_table[(x1, y1)] = [np.random.uniform(-5, 0) for i in range(4)]
if start_q_table is not None:
print("Loaded q-table")
with open(start_q_table, "rb") as f:
q_table = pickle.load(f)
episode_rewards = []
episode_scores = []
for episode in range(HM_EPISODES):
player = Snake(0)
enemy = Snake(1)
food = Food([player, enemy])
show = False
if episode % SHOW_EVERY == 0:
print(f"on #{episode}, epsilon is {epsilon}")
print(
f"{SHOW_EVERY} ep mean: {np.mean(episode_rewards[-SHOW_EVERY:])}"
)
print(
f"{SHOW_EVERY} ep mean score: {np.mean(episode_scores[-SHOW_EVERY:])}"
)
print(episode_scores)
else:
show = False
episode_reward = 0
episode_score = 0
def setUp(self):
print("start test")
self.butter = Food(4, 2, 1)