def test_min_reward(self):
print('test minimum reward')
env = SimpleVehicle()
state = env.reset()
state_size = env.state_size
action_size = env.discrete_action_size
agent = ReinforceAgent(state_size, action_size)
action = agent.getAction(state)
next_state, reward, done = env.discrete_step(action)
print('action: ', action)
print('reward: ', reward)
self.assertGreater(reward,0.001, 'reward is greater than 0.001')
def test_next_state_left_fast(self):
print('test left move fast')
env = SimpleVehicle()
next_state = env.reset()
print(next_state)
init = copy.copy(next_state)
action = 1# move left
next_state, reward, done = env.discrete_step(action)
print(next_state)
print(init)
self.assertNotEqual(init[0],next_state[0], 'initial and next state x-position are the same')
self.assertNotEqual(init[1],next_state[1], 'initial and next state y-position are the same')
self.assertLess(init[2],next_state[2], 'next state heading is less than initial')
def test_next_state_straight_fast(self):
print('test straight move fast')
env = SimpleVehicle()
next_state = env.reset()
print(next_state)
init = copy.copy(next_state)
action = 3# move straight fast
next_state, reward, done = env.discrete_step(action)
print(next_state)
print(init)
self.assertNotEqual(init[0],next_state[0], 'initial and next state x-position are the same')
self.assertNotEqual(init[1],next_state[1], 'initial and next state y-position are the same')
self.assertEqual(init[2],next_state[2], 'initial and next state heading are NOT the same')
def test_next_state_right(self):
print('test right move')
env = SimpleVehicle()
next_state = env.reset()
print(next_state)
init = copy.copy(next_state)
action = 4# move right
next_state, reward, done = env.discrete_step(action)
print(next_state)
print(init)
self.assertNotEqual(init[0],next_state[0], 'initial and next state x-position are the same')
self.assertNotEqual(init[1],next_state[1], 'initial and next state y-position are the same')
self.assertGreater(init[2],next_state[2], 'next state heading is greater than initial')
def test_done_condition(self):
print('check done condition')
env = SimpleVehicle()
state = env.reset()
state_size = env.state_size
action_size = env.discrete_action_size
agent = ReinforceAgent(state_size, action_size)
action = agent.getAction(state)
next_state, reward, done = env.discrete_step(action)
if (env._border_fcn(env.tmpstep[:, 0]) > env.tmpstep[:, 1]).any() \
or (env.tmpstep[:, 0] < env.xlim[0]).any() \
or (env.tmpstep[:, 0] > env.xlim[1]).any() \
or (env.tmpstep[:, 1] < env.ylim[0]).any() \
or (env.tmpstep[:, 1] > env.ylim[1]).any() \
or ((env.tmpstep[:, 0] > min(env.ox1)).any() and (env.tmpstep[:, 0] < max(env.ox1)).any() and (env.tmpstep[:, 1] > min(env.oy1)).any() and (env.tmpstep[:, 1] < max(env.oy1)).any()) \
or ((env.tmpstep[:, 0] > min(env.ox2)).any() and (env.tmpstep[:, 0] < max(env.ox2)).any() and (env.tmpstep[:, 1] > min(env.oy2)).any() and (env.tmpstep[:, 1] < max(env.oy2)).any()):
self.assertTrue(done, 'done is False')
else:
self.assertFalse(done, 'done is True')