def __init__(self, anchors, distance_model=None, edge_length=0.25):
"""
Initialise with a dictionary of anchors (id->(x,y), and other parameters)
"""
super(LocationEnginePDF, self).__init__()
self.anchors = anchors
self.edge_length = edge_length
if distance_model == None:
distance_model = DistanceModel()
self.distance_model = distance_model
self.pdfs = {}
# Find the range in x and y of the anchor positions
x, y = anchors.values()[0]
min_x, min_y = x, y
max_x, max_y = x, y
for anchor_id, location in anchors.iteritems():
x, y = location
min_x = min(x, min_x)
max_x = max(x, max_x)
min_y = min(y, min_y)
max_y = max(y, max_y)
self.grid = Grid(min_x - 1, max_x + 1, min_y - 1, max_y + 1)
self.combine_pdfs = self.multiply_pdfs
class LocationEnginePDF(object):
"""
Location Engine based on probability distributions
"""
def __init__(self, anchors, distance_model=None, edge_length=0.25):
"""
Initialise with a dictionary of anchors (id->(x,y), and other parameters)
"""
super(LocationEnginePDF, self).__init__()
self.anchors = anchors
self.edge_length = edge_length
if distance_model == None:
distance_model = DistanceModel()
self.distance_model = distance_model
self.pdfs = {}
# Find the range in x and y of the anchor positions
x, y = anchors.values()[0]
min_x, min_y = x, y
max_x, max_y = x, y
for anchor_id, location in anchors.iteritems():
x, y = location
min_x = min(x, min_x)
max_x = max(x, max_x)
min_y = min(y, min_y)
max_y = max(y, max_y)
self.grid = Grid(min_x - 1, max_x + 1, min_y - 1, max_y + 1)
self.combine_pdfs = self.multiply_pdfs
def set_standard(self):
self.distance_model = DistanceModel()
self.combine_pdfs = self.multiply_pdfs
self.pdfs = {}
def set_uniform(self):
self.distance_model = UniformDistanceModel()
self.combine_pdfs = self.add_pdfs
self.pdfs = {}
def generate_pdf(self, anchor_id, estimated_distance):
"A numpy array representing the probability of a tag being at each grid cell, given the distance from a particular base station."
ax, ay = self.anchors[anchor_id]
array = self.grid.array()
for ix, iy in self.grid.cell_indices():
x, y = self.grid.index_to_coordinate(ix, iy)
d = math.hypot(x - ax, y - ay)
array[ix][iy] = self.distance_model.distance_probability(d, estimated_distance)
return array
def round_distance(self, distance):
return float("%.1f" % distance)
def pdf(self, anchor_id, estimated_distance):
estimated_distance = self.round_distance(estimated_distance)
key = anchor_id, estimated_distance
if not self.pdfs.has_key(key):
self.pdfs[key] = self.generate_pdf(*key)
return self.pdfs[key]
def add_pdfs(self, distances):
""
pdf = self.grid.array(ones=False)
for anchor_id, estimated_distance in distances.iteritems():
pdf += self.pdf(anchor_id, estimated_distance)
return pdf
def multiply_pdfs(self, distances):
pdf = self.grid.array(ones=True)
for anchor_id, estimated_distance in distances.iteritems():
pdf *= self.pdf(anchor_id, estimated_distance)
return pdf
def coordinates(self, distances):
"""
The coordinates of the most likely tag location (x,y) that would give rise to the given distance measurements
Distances can either be a dictionary (base_id -> d) or a list of distances (in order of base id).
"""
pdf = self.combine_pdfs(distances)
i = pdf.argmax()
ix, iy = self.grid.divmod(i)
x, y = self.grid.index_to_coordinate(ix, iy)
return x, y
