def _getMaxActionValue(self, state):
maxActionId = 0
idAction = Action.IdToAction(maxActionId, state)
maxValue = self._getQValue(state, idAction)
for i in range(1, Action.getNumActions()):
idAction = Action.IdToAction(i, state)
currentValue = self._getQValue(state, idAction)
if currentValue > maxValue:
maxValue = currentValue
maxActionId = i
return maxActionId, maxValue
def loadQValues(self, filepath):
loadFile = open(filepath, 'r')
n = State.getNumStates()
m = Action.getNumActions()
for i in range(n):
for j in range(m):
self.qValues[i][j] = float(loadFile.readline())
loadFile.close()
def saveQValues(self, filepath):
saveFile = open(filepath, 'w')
n = State.getNumStates()
m = Action.getNumActions()
for i in range(n):
for j in range(m):
saveFile.write(str(self.qValues[i][j]) + "\n")
saveFile.close()
def __init__(self, initValue=0):
self.qValues = []
n = State.getNumStates()
m = Action.getNumActions()
for i in range(n):
# create state array of actions
self.qValues.append([])
# populate with default values
for j in range(m):
self.qValues[i].append(initValue)
def _getMaxActionValue(self, state):
numActions = Action.getNumActions()
stateTable = self.qValues[state.getStateId()]
maxActionId = 0
maxValue = self._getQValue(state, maxActionId)
for actionId in range(numActions):
currentValue = self._getQValue(state, actionId)
if currentValue > maxValue:
maxActionId = actionId
maxValue = currentValue
return maxActionId, maxValue
def getPiAction(self, state):
maxActionId, _ = self._getMaxActionValue(state)
maxAction = Action.IdToAction(maxActionId, state.convertToServoState())
return maxAction