def reset(self):
MazeTask.reset(self)
goUp = choice([True, False])
self.specialObs = goUp
if goUp:
self.env.goal = (3, self.length + 1)
else:
self.env.goal = (1, self.length + 1)
def reset(self):
MazeTask.reset(self)
goUp = choice([True, False])
self.specialObs = goUp
if goUp:
self.env.goal = (3, self.length + 1)
else:
self.env.goal = (1, self.length + 1)
def performAction(self, action):
poss = []
for a in range(self.actions):
if action - a % 4 != 2:
poss.append(a)
if random() < self.stochAction * len(poss):
MazeTask.performAction(self, choice(poss))
else:
MazeTask.performAction(self, action)
def performAction(self, action):
poss = []
for a in range(self.actions):
if action - a % 4 != 2:
poss.append(a)
if random() < self.stochAction * len(poss):
MazeTask.performAction(self, choice(poss))
else:
MazeTask.performAction(self, action)
def __init__(self, **args):
self.initPos = [(2, 1)]
self.setArgs(**args)
columns = [[1] * 5]
for dummy in range(self.length):
columns.append([1, 1, 0, 1, 1])
columns.append([1, 0, 0, 0, 1])
columns.append([1] * 5)
self.topology = array(columns).T
MazeTask.__init__(self, **args)
def __init__(self, **args):
self.initPos = [(2, 1)]
self.setArgs(**args)
columns = [[1] * 5]
for dummy in range(self.length):
columns.append([1, 1, 0, 1, 1])
columns.append([1, 0, 0, 0, 1])
columns.append([1] * 5)
self.topology = array(columns).T
MazeTask.__init__(self, **args)
def getReward(self):
if self.env.perseus[1] == self.length + 1:
if abs(self.env.perseus[0] - self.env.goal[0]) == 2:
# bad choice taken
self.env.perseus = self.env.goal
return self.bangPenalty
return MazeTask.getReward(self)
def getReward(self):
if self.env.perseus[1] == self.length + 1:
if abs(self.env.perseus[0] - self.env.goal[0]) == 2:
# bad choice taken
self.env.perseus = self.env.goal
return self.bangPenalty
return MazeTask.getReward(self)
def reset(self):
MazeTask.reset(self)
self.env.perseusDir = 1
def reset(self):
MazeTask.reset(self)
self.env.perseusDir = 1
def getReward(self):
if self.bad:
return self.minReward
else:
return MazeTask.getReward(self)
def reset(self):
MazeTask.reset(self)
self.bad = False
def getReward(self):
if self.bad:
return self.minReward
else:
return MazeTask.getReward(self)
def reset(self):
MazeTask.reset(self)
self.bad = False