pacman autograder : fail UCS test because of incorrectly testing if a node is a goal state before adding it into the queue

Working on Pacman UCS algorithm, the provided autograder PASS me on all tests case except goaldAtDequeue.test. Here is my code:

explored=[]
frontier=util.PriorityQueue()       #initiate Priority Queue
startState=(problem.getStartState(),'Stop',0)
frontier.push(startState,0)
explored.append(problem.getStartState())
parentPath=list()
childNode=list() # in the form of  [(x,y),'action1','action2','action3',]
while True:
    if frontier.isEmpty():
        return False         
    else:
        pNode=frontier.pop() 

    if pNode[0] == problem.getStartState():
        pass
    else:
        parentPath=[]
        parentPath=copy.deepcopy(pNode[1:])

    if problem.isGoalState(pNode[0]):
        return parentPath        


    for frontiers in problem.getSuccessors(pNode[0]):                       
            if frontiers[0] in explored:
                pass
            else:      
                childNode =[]                       #point childNode to an empty list
                childNode.append(frontiers[0])      #add successor's coordinate (x,y) into the childNode
                if parentPath!=[]:                  
                    childNode+=parentPath           #add the parent node 's path into the childNode
                childNode.append(frontiers[1])      #add the successors's move into the childNode

                pathCost=problem.getCostOfActions(childNode[1:])
               # print "current Child Node", frontiers[0]
              #  print " Path Cost", pathCost

                frontier.push(childNode,pathCost)            
                explored.append(frontiers[0])       #add state into explored list

The error autograder gave me was that:

Incorrectly testing if a node is a goal before adding it into the queue, instead of testing when you remove the node from the queue.