1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
|
'''
Author: Julian Ryde
'''
from __future__ import division
import time
import numpy as np
import mrol_mapping.mapper as mapper
import mrol_mapping.poseutil as poseutil
import mrol_mapping.mrol as mrol
import mrol_mapping.occupiedlist as occupiedlist
import mrol_mapping.pointcloud as pointcloud
import depth_image
import unittest
visualise = True
basedir = 'data/simulated/'
def load_poses(fname):
# skip comment lines
poses = {}
P = poseutil.Pose3D()
for line in open(fname):
if line.strip().startswith('#'):
continue
fields = line.split(None, 1)
fname = fields[0]
mat = np.asarray(eval(fields[1]))
P = poseutil.Pose3D()
# need to transpose from blender?
P.setMatrix(mat.T)
poses[fname] = P
return poses
true_poses = load_poses(basedir + 'poses.txt')
def get_PC(i):
'''Loads the i-th point cloud from test data and attach the true pose to
it'''
fname = str(i).rjust(4, '0') + '.png'
di = depth_image.load_image(basedir + fname)
pc = depth_image.image_to_points(di, max_range=di.max())
#pc = pointcloud.PointCloud(xyzs)
# remove ceiling and floor
# this could be done automatically with RANSAC
# large horizontal plane cause problems when contraining pose in
# horizontal plane?
# in noisy outdoor environments this is less of a problem?
pc.boxfilter(zmin=-1.5, zmax=0.5)
# Attach ground truth pose to the pointcloud
try:
pc.pose = true_poses[fname]
except KeyError as ke:
print ke
return pc, di
# TODO add benchmark test
class Testposeutil(unittest.TestCase):
def setUp(self):
#np.random.seed(6)
self.res = 0.04
self.vmap = mapper.VoxelMap(self.res, levels=3)
pc = get_PC(0)[0]
self.vmap.add_points(pc, pc.pose)
def check_localise(self, bestpose, truepose):
#print 'Best:', bestpose
#print 'True:', truepose
poseerror = truepose.getCombined(bestpose.inverse())
print 'Pose error:', poseerror
pos_error = np.sum(np.abs(poseerror.get()[:3]))
ang_error = np.sum(np.abs(poseerror.get()[3:]))
print '\nPosition error (m):', pos_error
print 'Angular error (degs):', np.degrees(ang_error)
position_tol = 0.04 # total position error tolerance
angular_tol = np.radians(0.55)
#self.assertTrue(pos_error < position_tol)
#self.assertTrue(ang_error < angular_tol)
#self.assertTrue(bestpose.is_close(truepose, position_tol, angular_tol))
def test_Ascanmatch_6DOF(self):
# "A" prepended to run this first cos it is quick
print "test_scanmatch_6DOF"
vmap = self.vmap
guessPose = get_PC(0)[0].pose
pc1 = get_PC(1)[0]
truepose = pc1.pose
start = time.time()
bestpose, maxcost = vmap.align_points(pc1, guessPose)
taken = time.time() - start
print 'Scanmatch time:', taken, 'seconds'
self.check_localise(bestpose, truepose)
def test_Ymapbuild_6DOF(self):
# "Z" prepended to run this last cos it is slow
print "test_mapbuild_6DOF"
vmap = self.vmap
guessPose = vmap.trajectory[0]
#guessPose = get_PC(0)[0].pose
if visualise:
vmap.display()
start = time.time()
for i in range(1, 16):
print 'Aligning scan', i
pc, di = get_PC(i)
bestpose, maxcost = vmap.align_points(pc, guessPose)
#bestpose = pc.pose # use the true value
print 'Bestpose:',bestpose
vmap.add_points(pc, bestpose)
self.check_localise(bestpose, pc.pose)
guessPose = bestpose
taken = time.time() - start
print 'Mapbuild time:', taken, 'seconds'
print 'Map size (voxels):', len(vmap)
print 'Perfect map has 19812 voxels, press "c" to continue.'
# TODO possibly interesting to calculate the ideal compression with the
# true poses, might be possible to get better than that?
def OFFtest_Zmapbuild_6DOF_feature(self):
# "Z" prepended to run this last cos it is slow
print "test_mapbuild_6DOF_feature"
#vmap = self.vmap
vmap = mapper.VoxelMap(self.res, levels=5)
pc, di = get_PC(0)
di_disconts = depth_image.depth_discont_sobel(di)
pc_disconts = depth_image.image_to_points(di_disconts)
vmap.add_points(pc_disconts, pc.pose)
guessPose = vmap.trajectory[0]
#guessPose = get_PC(0)[0].pose
if visualise:
import mrol_mapping.visualiser.robotvisualiser as robotvisualiser
vmap.display()
import pylab
pylab.ion()
pylab.imshow(di_disconts)
pylab.draw()
PC_vis = robotvisualiser.Visualiser(title='points')
PC_vis.setautominmax()
PC_vis.addmappts(pc_disconts.points)
start = time.time()
for i in range(1, 16):
print 'Aligning scan', i
pc, di = get_PC(i)
print "extracting feature set"
di_disconts = depth_image.depth_discont_sobel(di)
pc_disconts = depth_image.image_to_points(di_disconts)
if visualise:
pylab.imshow(di_disconts)
pylab.draw()
PC_vis.clear()
PC_vis.addmappts(pc_disconts.points)
print "performing alignment"
bestpose, maxcost = vmap.align_points(pc_disconts, guessPose, full_data=True)
#bestpose = pc.pose # use the true value
print 'Bestpose:',bestpose
#vmap.add_points(pc, bestpose)
vmap.add_points(pc_disconts, bestpose)
self.check_localise(bestpose, pc.pose)
guessPose = bestpose
taken = time.time() - start
print 'Mapbuild time:', taken, 'seconds'
print 'Map size (voxels):', len(vmap)
print 'Perfect map has 19812 voxels, press "c" to continue.'
# TODO possibly interesting to calculate the ideal compression with the
# true poses, might be possible to get better than that?
def OFFtest_saveload(self): # turned off because I run out of RAM and have to start paging.
print "test_saveload"
# TODO use tempfile module?
vmap = self.vmap
fname = '/tmp/vmap.map'
vmap.save(fname)
loaded = mapper.loadmap(fname)
# TODO assert equals needs __eq__ method on VoxelMap
def OFFtest_mapper_global_localisation(self):
dx = 0.5
dq = np.radians(15)
spread_pose = poseutil.Pose3D((dx, dx, 0, 0, 0, dq))
res = 0.04
map_mrol = mrol.MROL(res, levels=4)
#map_mrol.set_feature_range(8)
self.P1.volumetricsample(self.res+0.01111)
self.P2.volumetricsample(0.08)
map_mrol.addpoints(self.P1)
pc1 = get_PC(1)[0]
guesspose = pc1.pose
bestpose, maxo = map_mrol.match(pc1, guesspose, spread_pose)
truepose = self.P2.pose
if visualise:
self.P1.display(color=(1,0,0))
self.P2.pose = bestpose
self.P2.display(color=(0,1,0))
print 'Best:', bestpose
print 'True:', truepose
poseerror = truepose.getCombined(bestpose.inverse())
pos_error = np.sum(np.abs(poseerror.get()[:3]))
ang_error = np.sum(np.abs(poseerror.get()[3:]))
print
print 'Total Position error (m): ', pos_error
print 'Total Angular error (degs):', np.degrees(ang_error)
angular_tol = np.radians(2)
position_tol = 0.05
self.assertTrue(bestpose.is_close(truepose, position_tol, angular_tol))
def test_Amapper_raytrace(self): # "A" prepended to run this first cos it is quick
print "test_Amapper_raytrace"
res = 0.04
vm = mapper.VoxelMap(res, levels=3)
xyzs = np.zeros([4, 3])
xyzs[:, 1] = np.cumsum(np.ones(4))*res
xyzs[3, :] = [0.5, 0.5, 0.5] # this one isn't ray traced out
pc = pointcloud.PointCloud(xyzs)
vm.add_points(pc, None)
pose = (0, 0, 0, 0, 0, np.pi/2.)
ray_occs_pc = vm.ray_trace(pose, 50*res)
self.assertEqual(ray_occs_pc, pointcloud.PointCloud(occupiedlist.pointstovoxels(xyzs[:3], res, vm.lattice)))
ray_occs_pc = vm.ray_trace_pts(xyzs[0,:],xyzs[2,:])
self.assertEqual(ray_occs_pc, pointcloud.PointCloud(occupiedlist.pointstovoxels(xyzs[:3], res, vm.lattice)))
ray_occs_pc = vm.ray_trace_pts(np.array([0,0,0]),xyzs[3,:])
self.assertEqual(ray_occs_pc, pointcloud.PointCloud(occupiedlist.pointstovoxels(xyzs[3,:], res, vm.lattice)))
ray_occs_pc = vm.ray_trace_pts(xyzs[2,:],xyzs[0,:])
self.assertEqual(ray_occs_pc, pointcloud.PointCloud(occupiedlist.pointstovoxels(xyzs[:3], res, vm.lattice)))
# TODO ray check perpendicular to a plane
# test the free-space tracing
pc_555 = pointcloud.PointCloud(xyzs[3,:])
free_555_pc = vm.free_space_ray_trace(pc_555,np.asarray([0,0,0]),resolution=res)
xyzs_free_555 = np.array( [[ 0. , 0. , 0. ],
[ 0.04, 0.04, 0.04],
[ 0.08, 0.08, 0.08],
[ 0.12, 0.12, 0.12],
[ 0.16, 0.16, 0.16],
[ 0.2 , 0.2 , 0.2 ],
[ 0.24, 0.24, 0.24],
[ 0.28, 0.28, 0.28],
[ 0.32, 0.32, 0.32],
[ 0.36, 0.36, 0.36],
[ 0.4 , 0.4 , 0.4 ],
[ 0.44, 0.44, 0.44]] )
expected_555_pc = pointcloud.PointCloud(xyzs_free_555)
self.assertEqual(expected_555_pc, free_555_pc)
if __name__ == '__main__':
unittest.main()
|