def get_action(self, observation, reward = 0):
monsters = hf.get_monsters(observation)
mario = hf.get_mario(monsters)
state_arr = hf.getOkolica(observation,4,4,4,4)
state = state_arr.tostring()
action = self.get_q_action(state)
self.propagate_reward(reward)
if self.debug: self.print_world(state, state_arr)
return action
def get_action(self, observation, reward = 0):
monsters = hf.get_monsters(observation)
mario = hf.get_mario(monsters)
state_arr = hf.getOkolica(observation,4,4,4,4)
state = state_arr.tostring()
if (state.find("V") != -1
and abs(state.find("V") - state.find("M")) <5):
f = min(state.find("V"), state.find("M"))
t = max(state.find("V"), state.find("M")) +1
state = state[f:t]
action = self.get_q_action(state)
self.propagate_reward(reward)
if self.debug: self.print_world(state, state_arr)
return action
def print_world(self, s = [], sa = [], ok=100):
global all_scores, q, trial_actions, state, state_arr, \
observation, mario, monsters
observation = self.last_observation
monsters = hf.get_monsters(observation)
mario = hf.get_mario(monsters)
all_cores = self.all_scores
trial_actions = self.trial_actions
if len(sa) > 0: state_arr = sa
if len(s) > 0: state = s
q = self.Q
print "--------------------------------------------------"
s = hf.getOkolica(observation,ok,ok,ok,ok)
print "step: %d reward: %.2f " % \
(self.trial_steps, self.trial_reward)
print "\n".join(["".join(i) for i in s])
print "x: %2.2f y: %2.2f q-len: %d " % \
(mario.x, mario.y, len(self.Q))
print ""
