Python使用Tkinter实现机器人走迷宫
这本是课程的一个作业研究搜索算法,当时研究了一下Tkinter,然后写了个很简单的机器人走迷宫的界面,并且使用了各种搜索算法来进行搜索,如下图:
使用A*寻找最优路径:
由于时间关系,不分析了,我自己贴代码吧。希望对一些也要用Tkinter的人有帮助。
from Tkinter import * from random import * import time import numpy as np import util class Directions: NORTH = 'North' SOUTH = 'South' EAST = 'East' WEST = 'West' # Detect elements in the map window = Tk() window.title('CityBusPlanner') window.resizable(0,0) width = 25 (x, y) = (22, 22) totalsteps = 0 buidings = [(0, 0), (1, 0), (2, 0), (3, 0), (7, 0), (8, 0), (11, 0), (12, 0), (13, 0), (17, 0), (18, 0), (21, 0), (21, 1), (2, 2), (5, 2), (8, 2), (9, 2), (12, 2), (14, 2), (15, 2), (16, 2), (17, 2), (21, 2), (2, 3), (4, 3), (5, 3), (7, 3), (8, 3), (11, 3), (17, 3), (18, 3), (19, 3), (2, 4), (4, 4), (5, 4), (8, 4), (9, 4), (14, 4), (15, 4),(17, 4), (18, 4), (19, 4), (0, 6), (2, 6), (4, 6), (7, 6), (8, 6), (11, 6), (12, 6), (14, 6), (15, 6),(16, 6), (18, 6), (19, 6), (2, 7), (5, 7), (21, 7), (0, 8), (2, 8), (11, 8), (14, 8), (15, 8), (17, 8), (18, 8), (21, 8), (4, 9), (5, 9), (7, 9), (9, 9), (11, 9), (14, 9), (21, 9), (2, 10), (7, 10), (14, 10), (17, 10), (19, 10), (0, 11), (2, 11), (4, 11), (5, 11), (7, 11), (8, 11), (9, 11), (11, 11), (12, 11), (14, 11), (15, 11), (16, 11), (17, 11), (18, 11), (19, 11), (0, 13), (2, 13), (3, 13), (5, 13), (7, 13), (8, 13), (9, 13), (14, 13), (17, 13), (18, 13), (21, 13), (2, 14), (3, 14), (5, 14), (7, 14),(9, 14), (12, 14), (14, 14), (15, 14), (17, 14), (18, 14), (21, 14), (2, 15), (3, 15), (5, 15), (7, 15), (9, 15), (12, 15), (15, 15), (19, 15), (21, 15), (0, 16), (21, 16), (0, 17), (3, 17), (5, 17), (7, 17),(9, 17), (11, 17), (14, 17), (15, 17), (17, 17), (18, 17), (21, 17), (2, 18), (3, 18), (5, 18), (7, 18),(9, 18), (11, 18), (14, 18), (17, 18), (18, 18), (3, 19), (5, 19), (7, 19), (9, 19), (11, 19), (12, 19), (14, 19), (17, 19), (18, 19), (0, 21), (1, 21), (2, 21), (5, 21), (6, 21), (9, 21), (10, 21), (11, 21), (12, 21), (15, 21), (16, 21), (18, 21), (19, 21), (21, 21)] walls = [(10, 0), (0, 12), (21, 12), (14, 21)] park = [(14, 0), (15, 0), (16, 0)] robotPos = (21, 12) view = Canvas(window, width=x * width, height=y * width) view.grid(row=0, column=0) searchMapButton = Button(window,text = 'search') searchMapButton.grid(row = 0,column = 1) robotView = Canvas(window,width=x * width, height=y * width) robotView.grid(row = 0,column = 2) def formatColor(r, g, b): return '#%02x%02x%02x' % (int(r * 255), int(g * 255), int(b * 255)) def cityMap(): global width, x, y, buidings,walls,park,robot for i in range(x): for j in range(y): view.create_rectangle( i * width, j * width, (i + 1) * width, (j + 1) * width, fill='white', outline='gray', width=1) for (i, j) in buidings: view.create_rectangle( i * width, j * width, (i + 1) * width, (j + 1) * width, fill='black', outline='gray', width=1) for (i,j) in walls: view.create_rectangle( i * width, j * width, (i + 1) * width, (j + 1) * width, fill='blue', outline='gray', width=1) for (i,j) in park: view.create_rectangle( i * width, j * width, (i + 1) * width, (j + 1) * width, fill='red', outline='gray', width=1) def robotCityMap(): global width, x, y, buidings,walls,park,robot,robotPos for i in range(x): for j in range(y): robotView.create_rectangle( i * width, j * width, (i + 1) * width, (j + 1) * width, fill='black', width=1) robotView.create_rectangle( robotPos[0] * width, robotPos[1] * width, (robotPos[0] + 1) * width, (robotPos[1] + 1) * width, fill='white', outline='gray', width=1) # Create City Map cityMap() # Create Robot View robotCityMap() # Create a robot robot = view.create_rectangle(robotPos[0] * width + width * 2 / 10, robotPos[1] * width + width * 2 / 10, robotPos[0] * width + width * 8 / 10, robotPos[1] * width + width * 8 / 10, fill="orange", width=1, tag="robot") robotSelf = robotView.create_rectangle(robotPos[0] * width + width * 2 / 10, robotPos[1] * width + width * 2 / 10, robotPos[0] * width + width * 8 / 10, robotPos[1] * width + width * 8 / 10, fill="orange", width=1, tag="robot") visited = [robotPos] def move(dx,dy): global robot,x,y,robotPos,robotSelf,view global totalsteps totalsteps = totalsteps + 1 newX = robotPos[0] + dx newY = robotPos[1] + dy if (not isEdge(newX, newY)) and (not isBlock(newX, newY)): #print "move %d" % totalsteps view.coords(robot, (newX) * width + width * 2 / 10, (newY) * width + width * 2 / 10, (newX) * width + width * 8 / 10, (newY) * width + width * 8 / 10) robotView.coords(robotSelf, (newX) * width + width * 2 / 10, (newY) * width + width * 2 / 10, (newX) * width + width * 8 / 10, (newY) * width + width * 8 / 10) robotPos = (newX, newY) if robotPos not in visited: visited.append(robotPos) visitedPanel = robotView.create_rectangle( robotPos[0] * width, robotPos[1] * width, (robotPos[0] + 1) * width, (robotPos[1] + 1) * width, fill='white', outline='gray', width=1) robotView.tag_lower(visitedPanel,robotSelf) else: print "move error" def callUp(event): move(0,-1) def callDown(event): move(0, 1) def callLeft(event): move(-1, 0) def callRight(event): move(1, 0) def isBlock(newX,newY): global buidings,x,y for (i,j) in buidings: if (i == newX) and (j == newY): return True return False def isEdge(newX,newY): global x,y if newX >= x or newY >= y or newX < 0 or newY < 0 : return True return False def getSuccessors(robotPos): n = Directions.NORTH w = Directions.WEST s = Directions.SOUTH e = Directions.EAST successors = [] posX = robotPos[0] posY = robotPos[1] if not isBlock(posX - 1, posY) and not isEdge(posX - 1,posY): successors.append(w) if not isBlock(posX, posY + 1) and not isEdge(posX,posY + 1): successors.append(s) if not isBlock(posX + 1, posY) and not isEdge(posX + 1,posY): successors.append(e) if not isBlock(posX, posY -1) and not isEdge(posX,posY - 1): successors.append(n) return successors def getNewPostion(position,action): posX = position[0] posY = position[1] n = Directions.NORTH w = Directions.WEST s = Directions.SOUTH e = Directions.EAST if action == n: return (posX,posY - 1) elif action == w: return (posX - 1,posY) elif action == s: return (posX,posY + 1) elif action == e: return (posX + 1,posY) delay = False def runAction(actions): global delay n = Directions.NORTH w = Directions.WEST s = Directions.SOUTH e = Directions.EAST for i in actions: if delay: time.sleep(0.05) if i == n: #print "North" move(0, -1) elif i == w: #print "West" move(-1, 0) elif i == s: #print "South" move(0, 1) elif i == e: #sprint "East" move(1, 0) view.update() def searchMapTest(event): global robotPos actions = [] position = robotPos for i in range(100): successors = getSuccessors(position) successor = successors[randint(0, len(successors) - 1)] actions.append(successor) position = getNewPostion(position, successor) print actions runAction(actions) def reverseSuccessor(successor): n = Directions.NORTH w = Directions.WEST s = Directions.SOUTH e = Directions.EAST if successor == n: return s elif successor == w: return e elif successor == s: return n elif successor == e: return w roads = set() detectedBuildings = {} blockColors = {} blockIndex = 0 def updateBuildings(detectedBuildings): global robotView,width for block,buildings in detectedBuildings.items(): color = blockColors[block] for building in buildings: robotView.create_rectangle( building[0] * width, building[1] * width, (building[0] + 1) * width, (building[1] + 1) * width, fill=color, outline=color, width=1) def addBuilding(position): global blockIndex,detectedBuildings isAdd = False addBlock = '' for block,buildings in detectedBuildings.items(): for building in buildings: if building == position: return if util.manhattanDistance(position, building) == 1: if not isAdd: buildings.add(position) isAdd = True addBlock = block break else: #merge two block for building in detectedBuildings[block]: detectedBuildings[addBlock].add(building) detectedBuildings.pop(block) if not isAdd: newBlock = set([position]) blockIndex = blockIndex + 1 detectedBuildings['Block %d' % blockIndex] = newBlock color = formatColor(random(), random(), random()) blockColors['Block %d' % blockIndex] = color updateBuildings(detectedBuildings) def addRoad(position): global robotView,width,robotSelf visitedPanel = robotView.create_rectangle( position[0] * width, position[1] * width, (position[0] + 1) * width, (position[1] + 1) * width, fill='white', outline='gray', width=1) robotView.tag_lower(visitedPanel,robotSelf) def showPath(positionA,positionB,path): global robotView,width,view view.create_oval(positionA[0] * width + width * 3 / 10, positionA[1] * width + width * 3 / 10, positionA[0] * width + width * 7 / 10, positionA[1] * width + width * 7 / 10, fill='yellow', width=1) nextPosition = positionA for action in path: nextPosition = getNewPostion(nextPosition, action) view.create_oval(nextPosition[0] * width + width * 4 / 10, nextPosition[1] * width + width * 4 / 10, nextPosition[0] * width + width * 6 / 10, nextPosition[1] * width + width * 6 / 10, fill='yellow', width=1) view.create_oval(positionB[0] * width + width * 3 / 10, positionB[1] * width + width * 3 / 10, positionB[0] * width + width * 7 / 10, positionB[1] * width + width * 7 / 10, fill='yellow', width=1) hasDetected = set() def detectLocation(position): if position not in hasDetected: hasDetected.add(position) if isBlock(position[0],position[1]): addBuilding(position) elif not isEdge(position[0],position[1]): addRoad(position) def detect(position): posX = position[0] posY = position[1] detectLocation((posX,posY + 1)) detectLocation((posX,posY - 1)) detectLocation((posX + 1,posY)) detectLocation((posX - 1,posY)) def heuristic(positionA,positionB): return util.manhattanDistance(positionA,positionB) def AstarSearch(positionA,positionB): "Step 1: define closed: a set" closed = set() "Step 2: define fringe: a PriorityQueue " fringe = util.PriorityQueue() "Step 3: insert initial node to fringe" "Construct node to be a tuple (location,actions)" initialNode = (positionA,[]) initCost = 0 + heuristic(initialNode[0],positionB) fringe.push(initialNode,initCost) "Step 4: Loop to do search" while not fringe.isEmpty(): node = fringe.pop() if node[0] == positionB: return node[1] if node[0] not in closed: closed.add(node[0]) for successor in getSuccessors(node[0]): actions = list(node[1]) actions.append(successor) newPosition = getNewPostion(node[0], successor) childNode = (newPosition,actions) cost = len(actions) + heuristic(childNode[0],positionB) fringe.push(childNode,cost) return [] def AstarSearchBetweenbuildings(building1,building2): "Step 1: define closed: a set" closed = set() "Step 2: define fringe: a PriorityQueue " fringe = util.PriorityQueue() "Step 3: insert initial node to fringe" "Construct node to be a tuple (location,actions)" initialNode = (building1,[]) initCost = 0 + heuristic(initialNode[0],building2) fringe.push(initialNode,initCost) "Step 4: Loop to do search" while not fringe.isEmpty(): node = fringe.pop() if util.manhattanDistance(node[0],building2) == 1: return node[1] if node[0] not in closed: closed.add(node[0]) for successor in getSuccessors(node[0]): actions = list(node[1]) actions.append(successor) newPosition = getNewPostion(node[0], successor) childNode = (newPosition,actions) cost = len(actions) + heuristic(childNode[0],building2) fringe.push(childNode,cost) return [] def calculatePositions(buildingA,path): positions = set() positions.add(buildingA) nextPosition = buildingA for action in path: nextPosition = getNewPostion(nextPosition, action) positions.add(nextPosition) return positions def showRoad(fullRoad): global view,width for road in fullRoad: view.create_oval(road[0] * width + width * 4 / 10, road[1] * width + width * 4 / 10, road[0] * width + width * 6 / 10, road[1] * width + width * 6 / 10, fill='yellow', width=1) view.update() def search(node): successors = getSuccessors(node[0]) detect(node[0]) for successor in successors: nextPosition = getNewPostion(node[0], successor) if nextPosition not in roads: runAction([successor]) # to the next node roads.add(nextPosition) search((nextPosition,[successor],[reverseSuccessor(successor)])) runAction(node[2]) #back to top node def searchConsiderTopVisit(node,topWillVisit): successors = getSuccessors(node[0]) detect(node[0]) newTopWillVisit = set(topWillVisit) for successor in successors: nextPosition = getNewPostion(node[0], successor) newTopWillVisit.add(nextPosition) for successor in successors: nextPosition = getNewPostion(node[0], successor) if nextPosition not in roads and nextPosition not in topWillVisit: runAction([successor]) # to the next node roads.add(nextPosition) newTopWillVisit.remove(nextPosition) searchConsiderTopVisit((nextPosition,[successor],[reverseSuccessor(successor)]),newTopWillVisit) runAction(node[2]) #back to top node def searchShortestPathBetweenBlocks(block1,block2): shortestPath = [] buildingA = (0,0) buildingB = (0,0) for building1 in block1: for building2 in block2: path = AstarSearchBetweenbuildings(building1, building2) if len(shortestPath) == 0: shortestPath = path buildingA = building1 buildingB = building2 elif len(path) < len(shortestPath): shortestPath = path buildingA = building1 buildingB = building2 return (buildingA,buildingB,shortestPath) def addBuildingToBlocks(linkedBlock,buildingA): global detectedBuildings newLinkedBlock = linkedBlock.copy() for block,buildings in detectedBuildings.items(): for building in buildings: if util.manhattanDistance(buildingA, building) == 1: newLinkedBlock[block] = buildings break return newLinkedBlock def bfsSearchNextBlock(initBuilding,linkedBlock): global detectedBuildings closed = set() fringe = util.Queue() initNode = (initBuilding,[]) fringe.push(initNode) while not fringe.isEmpty(): node = fringe.pop() newLinkedBlock = addBuildingToBlocks(linkedBlock,node[0]) if len(newLinkedBlock) == len(detectedBuildings): return node[1] if len(newLinkedBlock) > len(linkedBlock): # find a new block actions = list(node[1]) ''' if len(node[1]) > 0: lastAction = node[1][len(node[1]) - 1] for successor in getSuccessors(node[0]): if successor == lastAction: nextPosition = getNewPostion(node[0], successor) actions.append(successor) return actions + bfsSearchNextBlock(nextPosition, newLinkedBlock) ''' return node[1] + bfsSearchNextBlock(node[0], newLinkedBlock) if node[0] not in closed: closed.add(node[0]) for successor in getSuccessors(node[0]): actions = list(node[1]) actions.append(successor) nextPosition = getNewPostion(node[0], successor) childNode = (nextPosition,actions) fringe.push(childNode) return [] def isGoal(node): global detectedBuildings,robotPos linkedBlock = {} positions = calculatePositions(robotPos, node[1]) for position in positions: for block,buildings in detectedBuildings.items(): for building in buildings: if util.manhattanDistance(position, building) == 1: linkedBlock[block] = buildings print len(linkedBlock) if len(linkedBlock) == 17: return True else: return False def roadHeuristic(road): return 0 def AstarSearchRoad(): global robotPos,detectedBuildings "Step 1: define closed: a set" closed = set() "Step 2: define fringe: a PriorityQueue " fringe = util.PriorityQueue() "Step 3: insert initial node to fringe" "Construct node to be a tuple (location,actions)" initRoad = (robotPos,[]) initCost = 0 + roadHeuristic(initRoad) fringe.push(initRoad,initCost) "Step 4: Loop to do search" while not fringe.isEmpty(): node = fringe.pop() if isGoal(node): print len(closed) return node[1] if node[0] not in closed: closed.add(node[0]) for successor in getSuccessors(node[0]): actions = list(node[1]) actions.append(successor) newPosition = getNewPostion(node[0], successor) childNode = (newPosition,actions) cost = len(actions) + roadHeuristic(childNode) fringe.push(childNode,cost) return [] def searchRoad(building): global detectedBuildings,robotPos linkedBlock = {} initBuilding = building return bfsSearchNextBlock(initBuilding,linkedBlock) def searchShortestRoad(): shortestRoad = [] shortestPositions = set() for block,buildings in detectedBuildings.items(): for building in buildings: road = searchRoad(building) positions = calculatePositions(building, road) if len(shortestPositions) == 0 or len(positions) < len(shortestPositions): shortestRoad = road shortestPositions = positions print len(shortestPositions) showRoad(shortestPositions) def searchMap(event): print "Search Map" global robotPos,roads,detectedBuildings,delay actions = [] #roads = set()s #roads.add(robotPos) #fringe = util.Stack() initNode = (robotPos,[],[]) # (position,forwardActions,backwarsdActions) #fringe.push(initNode) roads.add(robotPos) search(initNode) #searchConsiderTopVisit(initNode, set()) print detectedBuildings print len(detectedBuildings) #path = AstarSearchBetweenbuildings((6,21), (2, 18)) #showPath((6,21),(2,18), path) ''' shortestRoad = set() for block1 in detectedBuildings.values(): roads = set() for block2 in detectedBuildings.values(): if not block1 == block2: (buildingA,buildingB,path) = searchShortestPathBetweenBlocks(block1, block2) #showPath(buildingA,buildingB,path) positions = calculatePositions(buildingA,buildingB,path) roads = roads | positions if len(shortestRoad) == 0 or len(roads) < len(shortestRoad): shortestRoad = roads print len(shortestRoad) showRoad(shortestRoad) ''' ''' block1 = detectedBuildings.values()[3] print block1 block2 = detectedBuildings.values()[5] print block2 (buildingA,buildingB,path) = searchShortestPathBetweenBlocks(block1, block2) print buildingA,buildingB,path showPath(buildingA,buildingB,path) block1 = detectedBuildings.values()[10] print block1 block2 = detectedBuildings.values()[20] print block2 (buildingA,buildingB,path) = searchShortestPathBetweenBlocks(block1, block2) print buildingA,buildingB,path showPath(buildingA,buildingB,path) ''' searchShortestRoad() ''' path = searchRoad() #path = AstarSearchRoad() positions = calculatePositions(robotPos, path) print len(positions) showRoad(positions) delay = True #runAction(path) ''' window.bind("<Up>", callUp) window.bind("<Down>", callDown) window.bind("<Right>", callRight) window.bind("<Left>", callLeft) window.bind("s", searchMap) searchMapButton.bind("<Button-1>",searchMap) window.mainloop()
下面的util.py使用的是加州伯克利的代码:
# util.py # ------- # Licensing Information: You are free to use or extend these projects for # educational purposes provided that (1) you do not distribute or publish # solutions, (2) you retain this notice, and (3) you provide clear # attribution to UC Berkeley, including a link to http://ai.berkeley.edu. # # Attribution Information: The Pacman AI projects were developed at UC Berkeley. # The core projects and autograders were primarily created by John DeNero # (denero@cs.berkeley.edu) and Dan Klein (klein@cs.berkeley.edu). # Student side autograding was added by Brad Miller, Nick Hay, and # Pieter Abbeel (pabbeel@cs.berkeley.edu). import sys import inspect import heapq, random """ Data structures useful for implementing SearchAgents """ class Stack: "A container with a last-in-first-out (LIFO) queuing policy." def __init__(self): self.list = [] def push(self,item): "Push 'item' onto the stack" self.list.append(item) def pop(self): "Pop the most recently pushed item from the stack" return self.list.pop() def isEmpty(self): "Returns true if the stack is empty" return len(self.list) == 0 class Queue: "A container with a first-in-first-out (FIFO) queuing policy." def __init__(self): self.list = [] def push(self,item): "Enqueue the 'item' into the queue" self.list.insert(0,item) def pop(self): """ Dequeue the earliest enqueued item still in the queue. This operation removes the item from the queue. """ return self.list.pop() def isEmpty(self): "Returns true if the queue is empty" return len(self.list) == 0 class PriorityQueue: """ Implements a priority queue data structure. Each inserted item has a priority associated with it and the client is usually interested in quick retrieval of the lowest-priority item in the queue. This data structure allows O(1) access to the lowest-priority item. Note that this PriorityQueue does not allow you to change the priority of an item. However, you may insert the same item multiple times with different priorities. """ def __init__(self): self.heap = [] self.count = 0 def push(self, item, priority): # FIXME: restored old behaviour to check against old results better # FIXED: restored to stable behaviour entry = (priority, self.count, item) # entry = (priority, item) heapq.heappush(self.heap, entry) self.count += 1 def pop(self): (_, _, item) = heapq.heappop(self.heap) # (_, item) = heapq.heappop(self.heap) return item def isEmpty(self): return len(self.heap) == 0 class PriorityQueueWithFunction(PriorityQueue): """ Implements a priority queue with the same push/pop signature of the Queue and the Stack classes. This is designed for drop-in replacement for those two classes. The caller has to provide a priority function, which extracts each item's priority. """ def __init__(self, priorityFunction): "priorityFunction (item) -> priority" self.priorityFunction = priorityFunction # store the priority function PriorityQueue.__init__(self) # super-class initializer def push(self, item): "Adds an item to the queue with priority from the priority function" PriorityQueue.push(self, item, self.priorityFunction(item)) def manhattanDistance( xy1, xy2 ): "Returns the Manhattan distance between points xy1 and xy2" return abs( xy1[0] - xy2[0] ) + abs( xy1[1] - xy2[1] )
以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持【听图阁-专注于Python设计】。