def simpleCurveWithAvoidPoint(start_range, goal_range, attractor_range, plot=False):
# fig, ax = plt.subplots()
start = unif(start_range[0], start_range[1])
goal = unif(goal_range[0], goal_range[1])
attractor_point = unif(attractor_range[0], attractor_range[1])
avoid_point = attractor_point + np.random.rand(2) * 0.05
distractor_1 = unif([0.1, 0.1], [0.9, 0.9])
distractor_2 = unif([0.1, 0.1], [0.9, 0.9])
ts = np.array([0.0, 0.5, 1.0])
relevant_keypoints = np.array([start, attractor_point, goal])
all_features = np.array([start, avoid_point, attractor_point, distractor_1, distractor_2, goal])
spline = CubicSpline(ts, relevant_keypoints)
true_xys = spline(np.linspace(0, 1, 100))
unif_xys = equally_space_curve(true_xys, 100)
if plot:
# noisy_xys = true_xys + np.random.randn(100, 2) * 0.02
#ax.scatter(true_xys[:, 0], true_xys[:, 1], alpha=0.4, label='squashed')
# ax.scatter(noisy_xys[:, 0], noisy_xys[:, 1], alpha=0.25)
plt.scatter(unif_xys[:, 0], unif_xys[:, 1], label='uniform', alpha=0.5)
plt.scatter(all_features[0:2, 0], all_features[0:2, 1], marker='x', c='r', s=40*2)
# ax.scatter(avoid_point[0], avoid_point[1], s=30**2, alpha=0.3, c='r')
# plt.add_artist(plt.Circle(avoid_point, radius=0.1, alpha=0.2, color='r'))
plt.xlabel("x")
plt.ylabel("y")
plt.tight_layout()
return all_features, unif_xys
def movingSinWave(start_range, goal_range, attractor_range, plot=False):
start = unif(start_range[0], start_range[1])
goal = unif(goal_range[0], goal_range[1])
attractor = unif(attractor_range[0], attractor_range[1])
distractor_1 = unif([0.1, 0.1], [0.9, 0.9])
distractor_2 = unif([0.1, 0.1], [0.9, 0.9])
ts = np.array([0.0, 0.5, 1.0])
relevant_keypoints = np.array([start, attractor, goal])
spline = CubicSpline(ts, relevant_keypoints)
true_xys = spline(np.linspace(0, 1, 100))
unif_xys = equally_space_curve(true_xys, 100)
sin_noise = np.sin(2 * np.linspace(0, np.pi * 2, len(unif_xys))) * 0.2
unif_xys[:, 1] += sin_noise
unif_xys = equally_space_curve(unif_xys, 100)
all_features = np.array([start, attractor, distractor_1, distractor_2, goal])
if plot:
plt.scatter(unif_xys[:, 0], unif_xys[:, 1], label='uniform', alpha=0.5)
plt.scatter(all_features[0:2, 0], all_features[0:2, 1], marker='x', c='r', s=40*2)
plt.scatter(all_features[-1, 0], all_features[-1, 1], marker='x', c='r', s=40*2)
plt.xlabel("x")
plt.ylabel("y")
plt.ylim(bottom=0, top=1)
plt.tight_layout()
return all_features, unif_xys
def simpleCurveWithTwoPatrols(start_range, goal_range, ar_1, ar_2, plot=False):
start = unif(start_range[0], start_range[1])
goal = unif(goal_range[0], goal_range[1])
attractor_1 = unif(ar_1[0], ar_1[1])
attractor_2 = unif(ar_2[0], ar_2[1])
distractor_1 = unif([0.1, 0.1], [0.9, 0.9])
distractor_2 = unif([0.1, 0.1], [0.9, 0.9])
ts = np.array([0.0, 0.25, 0.75, 1.0])
relevant_keypoints = np.array([start, attractor_1, attractor_2, goal])
spline = CubicSpline(ts, relevant_keypoints)
true_xys = spline(np.linspace(0, 1, 100))
unif_xys = equally_space_curve(true_xys, 100)
patrol_1 = unif_xys[25] + unif([-0.05, 0.05], [-0.1, 0.1])
patrol_2 = unif_xys[75] + unif([0.05, -0.05], [0.1, -0.1])
all_features = np.array([start, patrol_1, patrol_2,
attractor_1, attractor_2, distractor_1, distractor_2, goal])
if plot:
plt.scatter(unif_xys[:, 0], unif_xys[:, 1], label='uniform', alpha=0.5)
plt.scatter(all_features[0:3, 0], all_features[0:3, 1], marker='x', c='r', s=40*2)
plt.xlabel("x")
plt.ylabel("y")
plt.tight_layout()
return all_features, unif_xys
def agents_setup(self):
'''
Create N agents, with random positions and starting headings.
'''
N = int(self.density * self.height * self.width)
for i in range(1, N + 1):
x = self.width * self.random.random()
if self.random.random() < 1 / 2:
y = unif(0, self.height / 2)
direction = False # ->
else:
y = unif(self.height / 2, self.height)
direction = True # <-
pos = np.array([x, y])
agent = Pedestrian(i, self, (x, y), direction)
self.space.place_agent(agent, pos)
self.schedule.add(agent)
def vuelos(numero):
muestra = []
for _ in range(numero):
Y = np.array([g + unif(-e, e) for e, g in zip(perturbacion, goal)])
vuelo, acciones = simulador(Y, Ze, 30, 800, jac=jac_f)
for estado, accion in zip(vuelo, acciones):
nuevo_estado = funcion(estado)
muestra.append(np.concatenate((accion, nuevo_estado), axis=None))
matriz = np.reshape(muestra, (799 * numero, 22))
with open('tablaaaaaaa.csv', 'w', newline='', encoding='utf-8') as csvfile:
writer = csv.writer(csvfile)
writer.writerows(muestra)
def unevenSpeeds(start_range, goal_range, attractor_range, plot=False):
start = unif(start_range[0], start_range[1])
goal = unif(goal_range[0], goal_range[1])
attractor = unif(attractor_range[0], attractor_range[1])
distractor_1 = unif([0.1, 0.1], [0.9, 0.9])
distractor_2 = unif([0.1, 0.1], [0.9, 0.9])
ts = np.array([0.0, 0.5, 1.0])
relevant_keypoints = np.array([start, attractor, goal])
spline = CubicSpline(ts, relevant_keypoints)
true_xys = spline(np.linspace(0, 1, 100))
unif_xys = equally_space_curve(true_xys, 100)
uneven_xys = np.concatenate((
equally_space_curve(unif_xys[0:30], 40),
equally_space_curve(unif_xys[30:70], 20),
equally_space_curve(unif_xys[70:], 40)
))
# print("Diffs: ", np.linalg.norm(uneven_xys[:-1] - uneven_xys[1:], axis=1))
# unif_xys = equally_space_curve(unif_xys, 100)
all_features = np.array([start, attractor, distractor_1, distractor_2, goal])
if plot:
# plt.scatter(unif_xys[:, 0], unif_xys[:, 1], label='uniform', alpha=0.5)
plt.scatter(uneven_xys[:, 0], uneven_xys[:, 1], label='uniform', alpha=0.5)
plt.scatter(all_features[0:2, 0], all_features[0:2, 1], marker='x', c='r', s=40*2)
plt.scatter(all_features[-1, 0], all_features[-1, 1], marker='x', c='r', s=40*2)
plt.xlabel("x")
plt.ylabel("y")
plt.ylim(bottom=0, top=1)
plt.tight_layout()
return all_features, uneven_xys
def randomize(xyz, missingness, extraness, random_extraness, stddev):
to_return = []
xyz = np.array(xyz)
to_return.append(np.random.normal(xyz, stddev))
# Extraness
if np.random.binomial(1, 1 - extraness) == 0:
for x in range(np.random.poisson(1)):
to_return.append(np.random.normal(xyz, stddev))
# Random Extraness
if np.random.binomial(1, 1 - random_extraness) == 0:
for x in range(np.random.poisson(1)):
to_return.append(
np.array([
unif(low=x_range[0], high=x_range[1]),
unif(low=y_range[0], high=y_range[1]),
unif(low=z_range[0], high=z_range[1])
]))
# Missingness
to_return = [] if (np.random.binomial(1, 1 - missingness)
== 0) else to_return
return np.array(to_return)
def data(self):
"""[summary]
Raises:
ValueError -- [if number of variable is less than 3, insuffieint to test MMI]
Returns:
[np array] -- [n_sample by n_variables matrix]
"""
if self.n_variables < 2 or self.n_samples < 1:
raise ValueError
else:
x = unif(self.low, self.high,
self.n_samples * (self.n_variables - 1)).reshape(
self.n_samples, self.n_variables - 1)
x = np.append(x,
np.remainder(np.sum(x, axis=1), 1)[:, None],
axis=1)
return x
demo_dir = "demos/splines2D-{}".format(t_stamp())
os.mkdir(demo_dir)
os.mkdir(demo_dir + "/train")
os.mkdir(demo_dir + "/val")
ts = np.array([0.0, 0.25, 0.75, 1.0])
fig, ax = plt.subplots()
# Training Data
for i in range(70):
start = unif([0.0, 0.0], [0.1, 0.1])
goal = unif([0.9, 0.9], [1.0, 1.0])
c1 = unif([0.5, 0.05], [0.65, 0.25])
c2 = unif([0.05, 0.5], [0.25, 0.65])
distractor_1 = unif([0.3, 0.3], [0.7, 0.7])
relevant_keypoints = np.array([start, c1, c2, goal])
all_keypoints = np.array([start, c1, c2, distractor_1, goal])
spline = CubicSpline(ts, relevant_keypoints)
true_xys = spline(np.linspace(0, 1, 100))
unif_xys = equally_space_curve(spline, 100)
# noisy_xys = true_xys + np.random.randn(100, 2) * 0.02
#ax.scatter(true_xys[:, 0], true_xys[:, 1], alpha=0.4, label='squashed')
# ax.scatter(noisy_xys[:, 0], noisy_xys[:, 1], alpha=0.25)
def reset(self):
x = self.width / 2 + unif(-2.5, 2.5)
y = self.height / 2 + unif(-2.5, 2.5)
phi = unif(0, 2 * pi)
return np.array([x, y, phi])
def reset(self):
self.i = 0
self.state = np.array(
[max(0, g + unif(-e, e)) for e, g in zip(self.p, self.goal)])
return self.state
