def get_fitness( self ):
if ( self._fitness != -1 ):
return self._fitness
else:
GeneticCode.surface.fill( (0,0,0) )
self.draw_onto_surface( GeneticCode.surface )
pixelArray = convertToPixelArray( GeneticCode.surface )
euclideanDist = euclideanDistance( pixelArray , IMG_PIXEL_ARRAY )
self._fitness = -(log10( euclideanDist ) - log10(norm( pixelArray ) + IMG_PIXEL_ARRAY_NORM ) )
return self._fitness
def get_fitness(self):
if (self._fitness != -1):
return self._fitness
else:
GeneticCode.surface.fill((0, 0, 0))
self.draw_onto_surface(GeneticCode.surface)
pixelArray = convertToPixelArray(GeneticCode.surface)
euclideanDist = euclideanDistance(pixelArray, IMG_PIXEL_ARRAY)
self._fitness = -(log10(euclideanDist) -
log10(norm(pixelArray) + IMG_PIXEL_ARRAY_NORM))
return self._fitness
def determine_dir(self):
v2center = self.vect_to_target(self.position)
d2center = norm(v2center)
ennemy_velocity = 1.3 * norm(
self._target[2:]) # Permet d'anticiper jusqu'à 30% de la vitesse max adverse connue
self.target_max_known_speed = max(self.target_max_known_speed, ennemy_velocity)
intercept_range_self = d2center * Jog.MAX_WHEEL_SPEED / self.target_max_known_speed
v2prerob = self.allies[0] - self.position
d2prerob = norm(v2prerob)
intercept_range_pre = d2prerob * Jog.MAX_WHEEL_SPEED / self.target_max_known_speed
v2postrob = self.allies[1] - self.position
d2postrob = norm(v2postrob)
intercept_range_post = d2postrob * Jog.MAX_WHEEL_SPEED / self.target_max_known_speed
p = d2prerob - intercept_range_pre - intercept_range_self
q = d2postrob - intercept_range_post - intercept_range_self
targ_dir = [0, 0]
# algorithmes de regulation des intercepteurs autour de l'ennemi
if (p < -Jog.SAFETY_MARGIN) and (q < -Jog.SAFETY_MARGIN):
targ_dir = v2center / d2center
elif p <= q and ((q > - Jog.SAFETY_MARGIN) != (p > -Jog.SAFETY_MARGIN)):
targ_dir = v2postrob / d2postrob
elif p > q and ((q > - Jog.SAFETY_MARGIN) != (p > -Jog.SAFETY_MARGIN)):
targ_dir = v2prerob / d2prerob
else:
if norm(self.vect_to_target(self.allies[1])) > d2center:
targ_dir = v2postrob / d2postrob
if norm(self.vect_to_target(self.allies[0])) < d2center:
targ_dir = v2prerob / d2prerob
return [Jog.NORMAL_SPEED, atan2(targ_dir[1], targ_dir[0])]
def bhattacharyya(x, y, Sx, Sy, K_factor=5):
BIG_L = TOM_SAPS.BIG_L
(_, m) = x.shape
(_, n) = y.shape
M = np.zeros((m, n))
dist = M.copy()
S = (Sx + Sy) / 2.0
Si = np.linalg.inv(S)
lnS = 0.5 * log(
np.linalg.det(S) / sqrt(np.linalg.det(Sx) * np.linalg.det(Sy)))
for i in range(m):
for j in range(n):
dx = x[:, i] - y[:, j]
dist[i, j] = norm(dx)
d = np.dot(np.dot(dx.T, Si), dx) ## NOTE: matrix multiplication
M[i, j] = d + lnS
if sqrt(d) > K_factor:
M[i, j] = BIG_L
return (M, dist)
''' Created on Sep 12, 2015 @author: mjchao ''' import pygame from Utils import convertToPixelArray, norm #list of global parameter constants P = 100 N = 1 K = 1 T = 500000 E = 500000 #---Change when image changes---# IMG = pygame.image.load( "../html/mona_lisa.bmp" ) OUTPUT_DIR = "tmp" IMG_WIDTH = 32 IMG_HEIGHT = 32 #--------------------------# IMG_PIXEL_ARRAY = convertToPixelArray( IMG ) IMG_PIXEL_ARRAY_NORM = norm( IMG_PIXEL_ARRAY )
