def test(ag1, ag2, render=False, load_model=False):
test_env = CMOTP()
if load_model:
ag1.load_model()
ag2.load_model()
ep_log = []
for iii in range(TEST_NUM):
print('test {}'.format(iii))
state1, state2 = test_env.reset()
ep_len = 0
while True:
if render:
test_env.render()
time.sleep(1)
action1 = ag1.choose_action(state1, 0, 0.0)
action2 = ag2.choose_action(state2, 0, 0.0)
next_state_, reward_, done_, _ = test_env.step([action1, action2])
next_state1, next_state2 = next_state_
state1 = next_state1
state2 = next_state2
ep_len += 1
if done_[0] or ep_len >= 1000:
ep_log.append(ep_len)
break
return np.mean(ep_log)
sts_2 = np.asarray(sts_2, dtype=np.float32).reshape(train_input_shape)
acts_2 = np.asarray(acts_2, dtype=np.int32).flatten()
rwds_2 = np.asarray(rwds_2, dtype=np.float32).flatten()
n_sts_2 = np.asarray(n_sts_2, dtype=np.float32).reshape(train_input_shape)
dns_2 = np.asarray(dns_2, dtype=np.bool).flatten()
ln_2 = np.asarray(ln_2, dtype=np.float32).flatten()
# train
agent1.train_without_replaybuffer(sts_1, acts_1, rwds_1, ln_1)
agent2.train_without_replaybuffer(sts_2, acts_2, rwds_2, ln_2)
else:
test_env = CMOTP()
agent1.load_model()
agent2.load_model()
for i in range(TEST_NUM):
state1, state2 = test_env.reset()
while True:
test_env.render()
time.sleep(1)
action1 = agent1.choose_action(state1, 0, 0.0)
action2 = agent2.choose_action(state2, 0, 0.0)
next_state, reward, done, _ = test_env.step([action1, action2])
next_state1, next_state2 = next_state
state1 = next_state1
state2 = next_state2
if done[0]:
break
env.close()
# next_state, reward, done, _ = env.step(action_n)
#
# len_this_episode += 1
# state = next_state
#
# if reward == 10.:
# break
# print(i, len_this_episode)
# len_episodes.append(len_this_episode)
#
# print(np.mean(len_episodes))
from Env.cmotp_IL import CMOTP
import time
if __name__ == '__main__':
env = CMOTP()
state = env.reset()
env.map = [[-1, -1, -1, 0, 0, 0, 0, 0, -1, -1, -1, -1],
[0, 0, 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, 0, 0, -1],
[-1, -1, -1, -1, 0, 0, 0, -1, -1, -1, -1, -1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1],
[0, 0, 0, -1, -1, -1, -1, -1, 0, 0, 0, -1],
[0, 0, 0, 0, 2, 3, 1, 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, 0, 0, -1],
[-1, -1, -1, -1, -1, 0, -1, -1, -1, -1, -1, -1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1]]
env.render()
time.sleep(100)