class Intersection: def __init__(self, length, prob, keepalive, wait_weight): self.NSRoad = Road(length,prob) self.EWRoad = Road(length,prob) self.keepalive = keepalive self.total_reward = 0 self.length = length self.prob = prob self.end = False self.wait_weight = wait_weight def getState(self): temp = np.array([np.array(self.NSRoad.binaryRepresentation()), np.array(self.EWRoad.binaryRepresentation())]) return temp def newInstance(self): self.NSRoad = Road(self.length, self.prob) self.EWRoad = Road(self.length, self.prob) self.end = False self.total_reward = 0 def getReward(self): return self.total_reward def gameEnd(self): return self.end def step(self,NS_action, EW_action): NSoutput = self.NSRoad.step(NS_action) EWoutput = self.EWRoad.step(EW_action) step_reward = 0 if (NSoutput[1][0] == 1) and (EWoutput[1][0] == 1): self.NSRoad.crash() self.EWRoad.crash() self.total_reward -= 100 self.end = True return -100 if (NSoutput[0][0] == 1): step_reward += 15 if (EWoutput[0][0] == 1): step_reward += 15 wait_penalty = self.wait_weight * (self.NSRoad.totalWait() + self.EWRoad.totalWait()) step_reward += (self.keepalive - wait_penalty) self.total_reward += step_reward return step_reward
class Intersection:
def __init__(self, length, prob, keepalive, wait_weight):
self.NSRoad = Road(length, prob)
self.EWRoad = Road(length, prob)
self.NSwait = 0
self.EWwait = 0
self.passedCars = 0
self.keepalive = keepalive
self.total_reward = 0
self.length = length
self.prob = prob
self.end = False
self.wait_weight = wait_weight
self.numSteps = 0
def getState(self):
temp = np.array([
self.NSRoad.binaryRepresentation(),
self.NSRoad.carRepresentation(),
self.EWRoad.binaryRepresentation(),
self.EWRoad.carRepresentation()
])
return temp
def getBinary(self):
state = []
temp = self.NSRoad.binaryRepresentation()
for i in range(0, 4):
state.append(temp[i])
temp = self.EWRoad.binaryRepresentation()
for i in range(0, 4):
state.append(temp[i])
state.append((self.NSwait + self.EWwait))
return np.array(state)
def newInstance(self):
self.NSRoad = Road(self.length, self.prob)
self.EWRoad = Road(self.length, self.prob)
self.end = False
self.passedCars = 0
self.total_reward = 0
def getReward(self):
return self.total_reward
def gameEnd(self):
return self.end
def gameWait(self):
print("Wait NS: " + str(self.NSRoad.totalWait()))
print("Wait EW: " + str(self.EWRoad.totalWait()))
def step(self, NS_action, EW_action):
if (NS_action == 1):
self.NSwait += 1
else:
self.NSwait = 0
if (EW_action == 1):
self.EWwait += 1
else:
self.EWwait = 0
#self.numSteps += 1
NSoutput = self.NSRoad.step(NS_action)
EWoutput = self.EWRoad.step(EW_action)
step_reward = 0
if ((NSoutput[1][0] == 1) and (EWoutput[1][0] == 1)) or (
(self.NSwait + self.EWwait) > 10) or (self.total_reward < -200):
self.NSRoad.crash()
self.EWRoad.crash()
self.total_reward -= 100
self.end = True
return -100
if (NSoutput[0][0] == 1):
step_reward += 20
self.passedCars += 1
if (EWoutput[0][0] == 1):
step_reward += 20
self.passedCars += 1
wait_penalty = self.wait_weight * (self.NSwait + self.EWwait)
step_reward += (self.keepalive - wait_penalty)
self.total_reward += step_reward
return step_reward