import json
import random
[docs]class Blurrer():
"""Simulate localization and vision noize.
Params is placed in the blurrer.json file.
Args:
object_angle_noize (float, optional): Noize for angle in radians.
Blurrer will uniformly random value from -object_angle_noize
to object_angle_noize and add it to the ground truth course.
Defaults to 0.
object_distance_noize (float, optional): Noize for distance in
percents divided by 100. Blurrer will uniformly random value
from -object_distance_noize to object_distance_noize and multiply
difference of 1 and this value with ground truth distance.
Defaults to 0..
observation_bonus (float, optional): Blurrer will increase the
consistency for every good observation (successfuly processed
image). Defaults to 0..
step_cost (float, optional): Blurrer will decrease the consistency
for every simulation step. Defaults to 0..
constant_loc_noize (float, optional): Constant localization noize.
Defaults to 0..
loc_noize_meters (float, optional): Multiplier for consistency, in
meters. Defaults to 0. Defaults to 0..
"""
def __init__(self, object_angle_noize=0., object_distance_noize=0.,
observation_bonus=0., step_cost=0.,
constant_loc_noize=0., loc_noize_meters=0.):
self.object_angle_noize = object_angle_noize
self.object_distance_noize = object_distance_noize
self.observation_bonus = observation_bonus
self.step_cost = step_cost
self.constant_loc_noize = constant_loc_noize
self.loc_noize_meters = loc_noize_meters
params = self.load_json("../player/blurrer.json")
self.consistency = 1
self.receiver = None
#penalty = self.receiver.player_state.penalty
[docs] def load_json(self, filename):
with open(filename) as f:
params = json.load(f)
self.object_angle_noize = params["object_angle_noize"]
self.object_distance_noize = params["object_distance_noize"]
self.observation_bonus = params["observation_bonus"]
self.step_cost = params["step_cost"]
self.constant_loc_noize = params["constant_loc_noize"]
self.loc_noize_meters = params["loc_noize_meters"]
[docs] def course(self, angle):
return angle + random.uniform(-self.object_angle_noize, self.object_angle_noize)
[docs] def distance(self, distance):
return distance * (1 + random.uniform(-self.object_distance_noize, self.object_distance_noize))
[docs] def objects(self, course=course, distance=distance):
return (self.course(course), self.distance(distance))
[docs] def loc(self, x, y):
return (self.coord(x), self.coord(y))
[docs] def coord(self, p):
random_factor = 1 - self.consistency
return p + self.loc_noize_meters * random.uniform(-random_factor, random_factor)
[docs] def step(self):
self.update_consistency(-self.step_cost)
[docs] def observation(self):
self.update_consistency(self.observation_bonus)
[docs] def update_consistency(self, value):
tmp = self.consistency + value
self.consistency = max(0, min(tmp, 1))