def __init__(self):
self.grid_width = 0.1
self.angle_blockwidth = np.pi / 8
self.env = car_sim_env()
self.state_action = []
self.tmp_state_action = []
self.cur_state = None
self.cur_action = None
def run(restore, LfD = False):
env = car_sim_env() # create environment (also adds some dummy traffic)
agt = env.create_agent(LearningAgent, test=False) # create agent
env.set_agent(agt, enforce_deadline=True) # specify agent to track
train_thread = threading.Thread(name="train", target=train, args=(env, agt, restore, LfD))
train_thread.daemon = True
train_thread.start()
env.plt_show()
while True:
continue
def run(restore):
env = car_sim_env()
agt = env.create_agent(LearningAgent, test=True)
env.set_agent(agt, enforce_deadline=False)
train_thread = threading.Thread(name="train",
target=train,
args=(env, agt, restore))
train_thread.daemon = True
train_thread.start()
env.plt_show()
while True:
continue
def run(restore):
env = car_sim_env()
agt = LearningAgent(env, is_test=False)
env.set_agent(agt, enforce_deadline=False)
#train_thread = threading.Thread(name="train", target=train, args=(env, agt, restore))
#train_thread.daemon = True # exit sub_threading when main_threading is done
#train_thread.start()
#plt_thread = threading.Thread(name="env.plt_show()", target=env.plt_show(), args=())
#plt_thread.daemon = True
#plt_thread.start()
env.plt_show()
train(env, agt, restore)
while True:
continue