def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
# Load the transition probabilities and store them in an integer-valued defaultdict.
# Use self.transProbDict[oldTile][newTile] to get the probability of transitioning from oldTile to newTile.
self.transProb = util.loadTransProb()
self.transProbDict = dict()
for (oldTile, newTile) in self.transProb:
if not oldTile in self.transProbDict:
self.transProbDict[oldTile] = collections.defaultdict(int)
self.transProbDict[oldTile][newTile] = self.transProb[(oldTile,
newTile)]
# Initialize the particles randomly.
self.particles = collections.defaultdict(int)
potentialParticles = list(self.transProbDict.keys())
for i in range(self.NUM_PARTICLES):
particleIndex = int(random.random() * len(potentialParticles))
self.particles[potentialParticles[particleIndex]] += 1
self.updateBelief()
def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
# Load the transition probabilities and store them in a dict of Counters
# self.transProbDict[oldTile][newTile] = probability of transitioning from oldTile to newTile
self.transProb = util.loadTransProb()
self.transProbDict = dict()
for (oldTile, newTile) in self.transProb:
if not oldTile in self.transProbDict:
self.transProbDict[oldTile] = collections.Counter()
self.transProbDict[oldTile][newTile] = \
self.transProb[(oldTile, newTile)]
# Initialize the particles randomly
self.particles = collections.Counter()
potentialParticles = self.transProbDict.keys()
for i in range(self.NUM_PARTICLES):
particleIndex = int(random.random() * len(potentialParticles))
self.particles[potentialParticles[particleIndex]] += 1
self.updateBelief()
def elapseTime(self):
if self.skipElapse: return ### ONLY FOR THE GRADER TO USE IN Problem 2
# BEGIN_YOUR_CODE (our solution is 7 lines of code, but don't worry if you deviate from this)
# Create the set of beliefs for the new time step
new_belief = util.Belief(self.belief.getNumRows(), self.belief.getNumCols(), value = 0.0)
for key in self.transProb.keys():
# Unpack old and new coordinates of tiles
old_x, old_y = key[0]
new_x, new_y = key[1]
# Unpack transition probability
transitionProb = self.transProb[key]
# Update probability
# Get the original from the current belief
oldProb = self.belief.getProb(old_x, old_y)
# Update with the current transition probability
newProb = oldProb*transitionProb
# Set the new probability in the new set of beliefs
# (we use addProb to sum as per the elimination formula)
new_belief.addProb(new_x, new_y, newProb)
self.belief = new_belief
# Re-normalize after updating
self.belief.normalize()
# In-place modification, return NoneType
return None
def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
# Load the transition probabilities and store them in an integer-valued defaultdict.
# Use self.transProbDict[oldTile][newTile] to get the probability of transitioning from oldTile to newTile.
self.transProb = util.loadTransProb()
self.transProbDict = dict()
for (oldTile, newTile) in self.transProb: # 遍历转移概率的keys
if not oldTile in self.transProbDict: # 同样,默认是指 键key 的集合.
self.transProbDict[oldTile] = collections.defaultdict(int)
self.transProbDict[oldTile][newTile] = self.transProb[(oldTile,
newTile)]
# 在构造 transProbDict - 它是一个字典的字典.（字典的嵌套）
# Initialize the particles randomly.
self.particles = collections.defaultdict(
int) # 索引是一个tile的位置(row,col), 对应的值是该位置上的粒子个数.
potentialParticles = list(
self.transProbDict.keys()) # 这时候存的是各个可能存在粒子的位置.
for i in range(self.NUM_PARTICLES):
particleIndex = int(random.random() * len(potentialParticles))
self.particles[potentialParticles[particleIndex]] += 1
self.updateBelief()
def observe(self, agentX, agentY, observedDist):
''' your code here'''
savedic = {}
for key in self.randomPart:
value = self.randomPart[key]
mean = math.sqrt((abs(agentX - util.colToX(key[1])))**2 +
(abs(agentY - util.rowToY(key[0])))**2)
pdf = util.pdf(mean, Const.SONAR_STD, observedDist)
newprob = value * pdf # prob in that tile and that tile's prob
savedic[key] = newprob
# pdf -> fill into savedic
count = 0
for key in self.realtemp:
count = count + 1
self.realtemp[key] = self.realtemp[key]
# random
self.randomPart = {} # initializing
for i in range(self.NUM_PARTICLES):
randIndex = util.weightedRandomChoice(savedic)
if not randIndex in self.randomPart:
self.randomPart[randIndex] = 1
else:
self.randomPart[randIndex] += 1
count = count + 1
NumRow = self.belief.getNumRows()
NumCol = self.belief.getNumCols()
new = util.Belief(NumRow, NumCol, 0)
for key in self.randomPart:
new.setProb(key[0], key[1], self.randomPart[key])
new.normalize()
self.belief = new
def __init__(self, numRows, numCols):
self.skipElapse = False ### ONLY USED BY GRADER.PY in case problem 3 has not been completed
# util.Belief is a class (constructor) that represents the belief for a single
# inference state of a single car (see util.py).
self.belief = util.Belief(numRows, numCols)
self.transProb = util.loadTransProb()
def updateBelief(self):
newBelief = util.Belief(self.belief.getNumRows(), self.belief.getNumCols(), 0)
for tile in self.particles:
newBelief.setProb(tile[0], tile[1], self.particles[tile])
newBelief.normalize()
self.belief = newBelief
def __init__(self, numRows, numCols):
self.skipElapse = False ### ONLY USED BY GRADER.PY in case problem 2 has not been completed
self.belief = util.Belief(numRows, numCols)
self.transProb = util.loadTransProb()
def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
''' initialize any variables you will need later '''
self.transprob = util.loadTransProb()
def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
def __init__(self, numRows, numCols):
self.belief = util.Belief(numRows, numCols)
''' initialize any variables you will need later '''
