def createTruth(npoints, nrepeats=4000):
"""
Generates a range of initial states for both vehicles, then simulates from
each one a certain number of times.
"""
## generating a grid of features
res = int(npoints**.25)
npoints = res**4
x_0 = np.linspace(-1., 1., res)
X_0 = np.empty((npoints, 4))
for feature in range(4):
X_0[:, feature] = np.repeat(np.tile(x_0, res**feature),
res**(3 - feature))
## randomly generating features
#X_0 = np.random.uniform(-1,1,size=(npoints,4))
y_0 = []
for point in range(npoints):
x = X_0[point, :]
mean1 = x[:2] * mean_bounds + mean_means
mean1 = motionModels.initialState_normal(mean1, noise * .1)
mean2 = x[2:4] * mean_bounds + mean_means
mean2 = motionModels.initialState_normal(mean2, noise * .1)
out = alarms.alarm_MCS(mean1, mean2, truthMM1, truthMM2, times,
nrepeats)
y_0 += [out[0]]
y_0 = np.array(y_0)
np.save(model_name + '_X.npy', X_0)
np.save(model_name + '_Y.npy', y_0)
v = np.random.normal(10., 5., nsims) - timeback * a
th = np.random.uniform(-np.pi, np.pi, nsims)
x = np.random.normal(0., 5., nsims) - timeback * v * np.cos(th)
y = np.random.normal(0., 5., nsims) - timeback * v * np.sin(th)
vehicle2 = np.array((x, y, th, v, a, w)).T
truth = []
results = defaultdict(list)
for sim in range(nsims):
veh1 = vehicle1[sim, :]
veh2 = vehicle2[sim, :]
veh1 = motionModels.initialState_normal(veh1, initialnoise)
veh2 = motionModels.initialState_normal(veh2, initialnoise)
# find 'real' collision occurrences by using very high-res particle alarm
pred, rt = alarms.alarm_MCS(veh1, veh2, MM1, MM2, times, 10000)
truth += [pred]
results['optimal'].append((pred, rt))
# estimators
starttime = timer()
result = alarms_nb.alarm_MCSbike(vehicle1[sim], initialnoise,
vehicle2[sim], initialnoise,
times.shape[0], timeres, noise, 10)
time = timer() - starttime
results['MC 10'].append((result, time))
starttime = timer()
result = alarms_nb.alarm_MCSbike(vehicle1[sim], initialnoise,
vehicle2[sim], initialnoise,
times.shape[0], timeres, noise, 100)
truth = []
results = defaultdict(list)
for sim in range(nsims):
veh1 = vehicle1[sim, :]
veh2 = vehicle2[sim, :]
veh1 = motionModels.initialState_normal(veh1, initialcovariance)
veh2 = motionModels.initialState_normal(veh2, initialcovariance)
# find 'true' collision occurrences by using very high-res MCS alarm
pred, rt = alarms.alarm_truth(veh1, veh2, MM1, MM2, np.array(times))
truth += [pred]
results['optimal'] += [(pred, rt)]
for nSamples in mc_samplecounts:
result = alarms.alarm_MCS(veh1, veh2, MM1, MM2, times, nSamples)
label = str(nSamples) + " MCS"
results[label] += [result]
result = model.alarm(veh1, veh2, MM1, MM2, times)
results['MLP'] += [result]
result = alarms.alarm_expected(veh1, veh2, MM1, MM2, times)
results['expected'] += [result]
result = alarms.alarm_UT_1(veh1, veh2, MM1, MM2, times)
results['UT 1'] += [result]
result = alarms.alarm_UT_2(veh1, veh2, MM1, MM2, times)
results['UT 2'] += [result]
v2 = np.random.normal(10., 5., nsims)
x2 = np.random.normal(-5, 2., nsims) - timeback * v2
vehicle2 = np.array((x2, v2)).T
print "running"
truth = []
results = defaultdict(list)
for sim in range(nsims):
veh1 = vehicle1[sim, :]
veh2 = vehicle2[sim, :]
veh1 = motionModels.initialState_normal(veh1, initialcovariance)
veh2 = motionModels.initialState_normal(veh2, initialcovariance)
# find 'true' collision occurrences by using very high-res MCS alarm
pred, rt = alarms.alarm_MCS(veh1, veh2, MM1, MM2, np.array(times), 10000)
truth += [pred]
results['optimal'].append((pred, rt))
# estimators
starttime = timer()
result = alarms_nb.alarm_MCSroad(vehicle1[sim], initialcovariance,
vehicle2[sim], initialcovariance,
times.shape[0], timeres, noise, 10)
time = timer() - starttime
results['MC 10'].append((result, time))
starttime = timer()
result = alarms_nb.alarm_MCSroad(vehicle1[sim], initialcovariance,
vehicle2[sim], initialcovariance,
times.shape[0], timeres, noise, 100)