Question 1：Reflex Agent
class ReflexAgent(Agent):
"""
A reflex agent chooses an action at each choice point by examining
its alternatives via a state evaluation function.
The code below is provided as a guide. You are welcome to change
it in any way you see fit, so long as you don't touch our method
headers.
"""
def getAction(self, gameState):
"""
You do not need to change this method, but you're welcome to.
getAction chooses among the best options according to the evaluation function.
Just like in the previous project, getAction takes a GameState and returns
some Directions.X for some X in the set {NORTH, SOUTH, WEST, EAST, STOP}
"""
legalMoves = gameState.getLegalActions()
scores = [self.evaluationFunction(gameState, action) for action in legalMoves]
bestScore = max(scores)
bestIndices = [index for index in range(len(scores)) if scores[index] == bestScore]
chosenIndex = random.choice(bestIndices)
"Add more of your code here if you want to"
return legalMoves[chosenIndex]
def evaluationFunction(self, currentGameState, action):
"""
Design a better evaluation function here.
The evaluation function takes in the current and proposed successor
GameStates (pacman.py) and returns a number, where higher numbers are better.
The code below extracts some useful information from the state, like the
remaining food (newFood) and Pacman position after moving (newPos).
newScaredTimes holds the number of moves that each ghost will remain
scared because of Pacman having eaten a power pellet.
Print out these variables to see what you're getting, then combine them
to create a masterful evaluation function.
"""
successorGameState = currentGameState.generatePacmanSuccessor(action)
newPos = successorGameState.getPacmanPosition()
newFood = successorGameState.getFood()
newGhostStates = successorGameState.getGhostStates()
newScaredTimes = [ghostState.scaredTimer for ghostState in newGhostStates]
"*** YOUR CODE HERE ***"
newFood = newFood.asList()
ghostPos = [(G.getPosition()[0], G.getPosition()[1]) for G in newGhostStates]
scared = min(newScaredTimes) > 0
if not scared and (newPos in ghostPos):
return -1.0
if newPos in currentGameState.getFood().asList():
return 1
closestFoodDist = sorted(newFood, key=lambda fDist: util.manhattanDistance(fDist, newPos))
closestGhostDist = sorted(ghostPos, key=lambda gDist: util.manhattanDistance(gDist, newPos))
fd = lambda fDis: util.manhattanDistance(fDis, newPos)
gd = lambda gDis: util.manhattanDistance(gDis, newPos)
return 1.0 / fd(closestFoodDist[0]) - 1.0 / gd(closestGhostDist[0])
Question 2：Minimax
class MinimaxAgent(MultiAgentSearchAgent):
"""
Your minimax agent (question 2)
"""
def getAction(self, gameState):
"""
Returns the minimax action from the current gameState using self.depth
and self.evaluationFunction.
Here are some method calls that might be useful when implementing minimax.
gameState.getLegalActions(agentIndex):
Returns a list of legal actions for an agent
agentIndex=0 means Pacman, ghosts are >= 1
gameState.generateSuccessor(agentIndex, action):
Returns the successor game state after an agent takes an action
gameState.getNumAgents():
Returns the total number of agents in the game
gameState.isWin():
Returns whether or not the game state is a winning state
gameState.isLose():
Returns whether or not the game state is a losing state
"""
"*** YOUR CODE HERE ***"
GhostIndex = [i for i in range(1, gameState.getNumAgents())]
def term(state, d):
return state.isWin() or state.isLose() or d == self.depth
def min_value(state, d, ghost):
if term(state, d):
return self.evaluationFunction(state)
"Value for Min node. May have multiple ghosts"
v = 10000000000000000
for action in state.getLegalActions(ghost):
if ghost == GhostIndex[-1]:
v = min(v, max_value(state.generateSuccessor(ghost, action), d + 1))
else:
v = min(v, min_value(state.generateSuccessor(ghost, action), d, ghost + 1))
return v
def max_value(state, d):
if term(state, d):
return self.evaluationFunction(state)
"Value for Max node"
v = -10000000000000000
for action in state.getLegalActions(0):
v = max(v, min_value(state.generateSuccessor(0, action), d, 1))
return v
"Select action for Max node"
res = [(action, min_value(gameState.generateSuccessor(0, action), 0, 1)) for action in
gameState.getLegalActions(0)]
res.sort(key=lambda k: k[1])
return res[-1][0]
util.raiseNotDefined()
Question 3：Alpha-Beta Pruning
class AlphaBetaAgent(MultiAgentSearchAgent):
"""
Your minimax agent with alpha-beta pruning (question 3)
"""
def getAction(self, gameState):
"""
Returns the minimax action using self.depth and self.evaluationFunction
"""
"*** YOUR CODE HERE ***"
GhostIndex = [i for i in range(1, gameState.getNumAgents())]
inf = 1e100
def term(state, d):
return state.isWin() or state.isLose() or d == self.depth
def min_value(state, d, ghost, A, B):
if term(state, d):
return self.evaluationFunction(state)
v = inf
for action in state.getLegalActions(ghost):
if ghost == GhostIndex[-1]:
v = min(v, max_value(state.generateSuccessor(ghost, action), d + 1, A, B))
else:
v = min(v, min_value(state.generateSuccessor(ghost, action), d, ghost + 1, A, B))
if v < A:
return v
B = min(B, v)
return v
def max_value(state, d, A, B):
if term(state, d):
return self.evaluationFunction(state)
v = -inf
for action in state.getLegalActions(0):
v = max(v, min_value(state.generateSuccessor(0, action), d, 1, A, B))
if v > B:
return v
A = max(A, v)
return v
def alphabeta(state):
v = -inf
act = None
A = -inf
B = inf
for action in state.getLegalActions(0):
tmp = min_value(gameState.generateSuccessor(0, action), 0, 1, A, B)
if v < tmp:
v = tmp
act = action
if v > B:
return v
A = max(A, tmp)
return act
return alphabeta(gameState)
util.raiseNotDefined()
Question 4：Expectimax
class ExpectimaxAgent(MultiAgentSearchAgent):
"""
Your expectimax agent (question 4)
"""
def getAction(self, gameState):
"""
Returns the expectimax action using self.depth and self.evaluationFunction
All ghosts should be modeled as choosing uniformly at random from their
legal moves.
"""
"*** YOUR CODE HERE ***"
GhostIndex = [i for i in range(1, gameState.getNumAgents())]
def term(state, d):
return state.isWin() or state.isLose() or d == self.depth
def exp_value(state, d, ghost):
if term(state, d):
return self.evaluationFunction(state)
v = 0
prob = 1 / len(state.getLegalActions(ghost))
for action in state.getLegalActions(ghost):
if ghost == GhostIndex[-1]:
v += prob * max_value(state.generateSuccessor(ghost, action), d + 1)
else:
v += prob * exp_value(state.generateSuccessor(ghost, action), d, ghost + 1)
return v
def max_value(state, d):
if term(state, d):
return self.evaluationFunction(state)
v = -10000000000000000
for action in state.getLegalActions(0):
v = max(v, exp_value(state.generateSuccessor(0, action), d, 1))
return v
res = [(action, exp_value(gameState.generateSuccessor(0, action), 0, 1)) for action in
gameState.getLegalActions(0)]
res.sort(key=lambda k: k[1])
return res[-1][0]
util.raiseNotDefined()
Question 5：Evaluation Function
def betterEvaluationFunction(currentGameState):
"""
Your extreme ghost-hunting, pellet-nabbing, food-gobbling, unstoppable
evaluation function (question 5).
DESCRIPTION:
"""
"*** YOUR CODE HERE ***"
newPos = currentGameState.getPacmanPosition()
newFood = currentGameState.getFood().asList()
newGhostStates = currentGameState.getGhostStates()
newScaredTimes = [ghostState.scaredTimer for ghostState in newGhostStates]
eval = currentGameState.getScore()
foodDist = float("inf")
for food in newFood:
foodDist = min(foodDist, util.manhattanDistance(food, newPos))
eval += 1.0 / foodDist
return eval
