crawler.py 文件源码

def doAction(self, action):
       """
         Perform the action and update
         the current state of the Environment
         and return the reward for the
         current state, the next state
         and the taken action.

         Returns:
           nextState, reward
       """
       nextState, reward =  None, None

       oldX,oldY = self.crawlingRobot.getRobotPosition()

       armBucket,handBucket = self.state
       armAngle,handAngle = self.crawlingRobot.getAngles()
       if action == 'arm-up':
         newArmAngle = self.armBuckets[armBucket+1]
         self.crawlingRobot.moveArm(newArmAngle)
         nextState = (armBucket+1,handBucket)
       if action == 'arm-down':
         newArmAngle = self.armBuckets[armBucket-1]
         self.crawlingRobot.moveArm(newArmAngle)
         nextState = (armBucket-1,handBucket)
       if action == 'hand-up':
         newHandAngle = self.handBuckets[handBucket+1]
         self.crawlingRobot.moveHand(newHandAngle)
         nextState = (armBucket,handBucket+1)
       if action == 'hand-down':
         newHandAngle = self.handBuckets[handBucket-1]
         self.crawlingRobot.moveHand(newHandAngle)
         nextState = (armBucket,handBucket-1)

       newX,newY = self.crawlingRobot.getRobotPosition()

       # a simple reward function
       reward = newX - oldX

       self.state = nextState
       return nextState, reward