def test_no_taste(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
observation = blobble_env.reset_test([3, 3, 5], TEST_FOOD, taste=False)
self.assertEqual(0, observation[3], 'Should be no taste 0')
observation, reward, done, _ = blobble_env.step(4) # Step West
self.assertEqual(0, observation[3], 'Should be no taste 0')
def test_move_east(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(3) # Step East
self.assertEqual(0, observation[0], 'Should be northerly=0')
self.assertEqual(1, observation[1], 'Should be easterly=1')
def test_move_too_far_north(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([10, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(1) # Step North
self.assertEqual(10, observation[0], 'Should be northerly=10')
self.assertEqual(0, observation[1], 'Should be easterly=0')
def test_move_too_far_west(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, -10, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(4) # Step West
self.assertEqual(0, observation[0], 'Should be northerly=0')
self.assertEqual(-10, observation[1], 'Should be easterly=-10')
def test_reward_eat_no_nutritional_value(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # Eat food=0
self.assertEqual(5, observation[2], 'Should be health=5')
self.assertEqual(0.5, reward, 'Should be reward=0.5')
def test_eat_no_step(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # No step
self.assertEqual(0, observation[0], 'Should be northerly=0')
self.assertEqual(0, observation[1], 'Should be easterly=0')
def test_eat(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
observation = blobble_env.reset_test([3, 3, 5], TEST_FOOD)
self.assertEqual(1, observation[3], 'Should be food of value 1')
observation, reward, done, _ = blobble_env.step(0) # Eat food
self.assertEqual(0, observation[3], 'Should be no food')
def test_move_south(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # No step
observation, reward, done, _ = blobble_env.step(1) # Step North
observation, reward, done, _ = blobble_env.step(2) # Step South
self.assertEqual(0, observation[0], 'Should be northerly=0')
self.assertEqual(0, observation[1], 'Should be easterly=0')
def test_reward_eat_nutritional_value(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(3) # Move East
observation, reward, done, _ = blobble_env.step(1) # Move North
observation, reward, done, _ = blobble_env.step(0) # Eat food=5
self.assertEqual(10, observation[2], 'Should be health=10')
self.assertEqual(1.0, reward, 'Should be reward=1.0')
def test_eat_and_move_west(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
observation = blobble_env.reset_test([3, 3, 5], TEST_FOOD)
print(observation)
self.assertEqual(1, observation[3], 'Should be food of value 1')
observation, reward, done, _ = blobble_env.step(8) # Eat food, Go West
print(observation)
observation, reward, done, _ = blobble_env.step(3) # Go East back to original (no food) spot
print(observation)
self.assertEqual(0, observation[3], 'Should be no food')
def test_smell_south(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # Eat food at current location
'''
[ ., ., B, ., .]
[ ., 2, 0, -5 .],
[ ., ., 0, ., .],
'''
self.assertEqual(-2, observation[5], 'Should be sniff south=0.75')
def test_smell_west(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # Eat food at current location
'''
[ ., 4, ., ., .],
[ 2, 4, B, ., .],
[ ., 2, ., ., .],
'''
self.assertEqual(0, observation[7], 'Should be sniff west=3')
def test_smell_north(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([0, 0, 5], TEST_FOOD)
observation, reward, done, _ = blobble_env.step(0) # Eat food at current location
'''
[ ., ., ., ., .],
[ ., 4, 2, 3, .],
[ ., ., B, ., .]
'''
self.assertEqual(3, observation[4], 'Should be sniff north=4')
def test_taste(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
observation = blobble_env.reset_test([0, 0, 5], TEST_FOOD)
self.assertEqual(0, observation[3], 'Should be taste=0')
observation = blobble_env.reset_test([-1, -4, 5], TEST_FOOD)
self.assertEqual(-2, observation[3], 'Should be taste=-2')
observation = blobble_env.reset_test([3, 3, 5], TEST_FOOD)
self.assertEqual(1, observation[3], 'Should be taste=1')
observation = blobble_env.reset_test([3, 4, 5], TEST_FOOD)
self.assertEqual(-2, observation[3], 'Should be taste=-2')
def test_no_smell(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
observation = blobble_env.reset_test([3, 3, 5], TEST_FOOD, smell=False)
self.assertEqual(0, observation[4], 'Should be no smell north')
self.assertEqual(0, observation[5], 'Should be no smell south')
self.assertEqual(0, observation[6], 'Should be no smell east')
self.assertEqual(0, observation[7], 'Should be no smell west')
observation, reward, done, _ = blobble_env.step(4) # Step West
self.assertEqual(0, observation[4], 'Should be no smell north after step')
self.assertEqual(0, observation[5], 'Should be no smell south after step')
self.assertEqual(0, observation[6], 'Should be no smell east after step')
self.assertEqual(0, observation[7], 'Should be no smell west after step')
def test_reset_observation(self):
blobble_env = BlobbleEnv()
observation = blobble_env.reset()
expected = np.array((0, 0, 5), dtype=np.float)
np.testing.assert_array_equal(expected, observation[0:3], 'Observation not as expected after reset')
observation = blobble_env.reset_test([1, 2, 3], TEST_FOOD)
expected = np.array((1, 2, 3), dtype=np.float)
np.testing.assert_array_equal(expected, observation[0:3], 'Observation not as expected after reset test')
def create_blobble_video(video_filename, num_episodes=1, fps=30):
filename = video_filename + ".mp4"
blobble_env = BlobbleEnv()
print('Observation space: '+str(blobble_env.observation_space))
print('Action space: '+str(blobble_env.action_space))
with imageio.get_writer(filename, fps=fps) as video:
for i in range(num_episodes-1):
print(i)
blobble_env.reset()
done = False
while not done:
action = np.random.randint(8)
print('Next Action: ', action)
observation, reward, done, _ = blobble_env.step(action)
blobble_env.render_print()
video.append_data(blobble_env.render(mode='rgb_array'))
blobble_env.close()
def test_render_human(self):
blobble_env = BlobbleEnv()
blobble_env.reset()
blobble_env.reset_test([-2, 5, 5], TEST_FOOD)
blobble_env.render(mode='human')