values = []
for i in np.arange(len(y)):
if(len(y[i]) > j):
values.append(y[i][j]-average)
means.append(mean(values))
# stds.append((mean(np.array(values)**(2.0)))**0.5) # TODO
stds.append(std(values))
plot(np.arange(200)*0.05,np.ones(200)*average,'b', lw=3)
plot(np.arange(200)*0.05,average+np.array(means),'r', lw=3)
plot(np.arange(200)*0.05,average+np.array(means)+np.array(stds),'r--', lw=3)
plot(np.arange(200)*0.05,average+np.array(means)-np.array(stds),'r--', lw=3)
if len(plotLeutiDistances) > 0: # Leute Score evaluation
spacings = plotLeutiDistances
if doRovio:
leutiOutputRovio = td_rovio.computeLeutiScore(rovio_posID, rovio_attID, rovio_velID, td_vicon, vicon_posID, vicon_attID, plotLeutiDistances, spacings, 0.0)
figure()
ax = axes()
boxplot(leutiOutputRovio, positions = np.arange(len(plotLeutiDistances)), widths = 0.8)
ax.set_xticklabels(plotLeutiDistances)
ax.set_xlabel('Travelled Distance [m]')
ax.set_ylabel('Position Error [m]')
if doOkvis:
leutiOutputOkvis = td_okvis.computeLeutiScore(okvis_posID, okvis_attID, okvis_velID, td_vicon, vicon_posID, vicon_attID, plotLeutiDistances, spacings, 0.0)
figure()
ax = axes()
boxplot(leutiOutputOkvis, positions = np.arange(len(plotLeutiDistances)), widths = 0.8)
ax.set_xticklabels(plotLeutiDistances)
ax.set_xlabel('Travelled Distance [m]')
ax.set_ylabel('Position Error [m]')
plot(x[i],
np.abs(np.array(y[i]) - average) / np.array(cov[i]),
'#999999')
plt.axis([0, plotFeaTimeEnd, 0, 5])
plt.legend()
plt.suptitle('Convergence of Landmark Distance')
axis.set_ylabel('Normalized error')
axis.set_xlabel('Time [s]')
if len(plotLeutiDistances) > 0: # Leute Score evaluation
spacings = plotLeutiDistances
if leutiSpacing > 0:
spacings = np.ones(len(plotLeutiDistances)) * leutiSpacing
if doRovio:
leutiOutputRovio = td_rovio.computeLeutiScore('pos', 'att', 'vel',
td_vicon, 'pos', 'att',
plotLeutiDistances,
spacings, 0.0)
figure()
ax = axes()
p = np.arange(len(plotLeutiDistances))
boxplot(leutiOutputRovio, positions=p, widths=0.8)
ax.set_xticklabels(plotLeutiDistances)
ax.set_xlabel('Travelled Distance [m]')
ax.set_ylabel('Position Error [m]')
med = np.zeros(len(leutiOutputRovio))
for i in range(len(leutiOutputRovio)):
med[i] = median(leutiOutputRovio[i])
s = np.array(plotLeutiDistances)**0.5
a = s.dot(med) / s.dot(s)
fit = a * s
plot(p, fit)
axis = subplot(2,1,2)
line_th = plot([0, plotFeaTimeEnd], [sqrt(6.64), sqrt(6.64)],'g--', lw=3, label='1% threshold')
for i in np.arange(len(x)):
plot(x[i],np.abs(np.array(y[i])-average)/np.array(cov[i]),'#999999')
plt.axis([0, plotFeaTimeEnd, 0, 5])
plt.legend()
plt.suptitle('Convergence of Landmark Distance')
axis.set_ylabel('Normalized error')
axis.set_xlabel('Time [s]')
if len(plotLeutiDistances) > 0: # Leute Score evaluation
spacings = plotLeutiDistances
if leutiSpacing > 0:
spacings = np.ones(len(plotLeutiDistances))*leutiSpacing
if doRovio:
leutiOutputRovio = td_rovio.computeLeutiScore('pos', 'att', 'vel', td_vicon, 'pos', 'att', plotLeutiDistances, spacings, 0.0)
figure()
ax = axes()
p = np.arange(len(plotLeutiDistances))
boxplot(leutiOutputRovio, positions = p, widths = 0.8)
ax.set_xticklabels(plotLeutiDistances)
ax.set_xlabel('Travelled Distance [m]')
ax.set_ylabel('Position Error [m]')
med = np.zeros(len(leutiOutputRovio))
for i in range(len(leutiOutputRovio)):
med[i] = median(leutiOutputRovio[i])
s = np.array(plotLeutiDistances)**0.5
a = s.dot(med)/s.dot(s)
fit = a*s
plot(p,fit)
title('Error Plot for Position, a = ' + str(a))
