-
Notifications
You must be signed in to change notification settings - Fork 2
/
test_gen_paper_results.m
279 lines (224 loc) · 9.14 KB
/
test_gen_paper_results.m
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
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
% This script does the following
clc, clear
run('/home/brucebot/workspace/test/CVPR17/setup.m')
addpath(genpath('/home/brucebot/workspace/lc-calibration/L1_relaxation'));
poolobj = gcp('nocreate');
rmpath(genpath('/home/brucebot/workspace/lc-calibration/LiDAR_intrinsic_calibration'))
% If no pool, do not create new one.
if isempty(poolobj)
parpool('eight-threading');
end
project_name = "OptimalShape";
clear_fig = 1;
save_fig = 0;
fig = createFigureOptions(10, 1, project_name, clear_fig, 1, 30);
[axes_handles, fig_handles] = createFigHandleWithOptions(fig);
disp("figure handles created")
% paths
opts.save_path = "./data/July-2021/";
opts.load_path = "./data/July-2021/";
opts.filename = "original_shape.mat";
% user parameters
opts.regen_LiDAR_points = 0; % 0: if file exist, use it, or regenerate it
opts.use_best_shape = 1;
opts.plot = 1;
opts.save_plots = 0;
% 0: simulation
% 1: lab
% 2: wavefield
% 3: robotics building
% 4: calibration (25 targets): BAD opts.target_num: 16, 21
% 5: old data (no images, 18 targets): GOOD opts.target_num = 2, 6, 7, 8, 9, 11, 12
% 6: vary distance 7/14/2021 (max 15)
% 7: calibration data 7/14/2021 (max 12)
opts.dataset = 0;
% In exp: which target to use.
% In sim, which list of angle and translation to use
% check getAngleNTranslationList for reference
% 1-99: no noise
% 101-199, level 1 noise
% 201-299, level 2 noise
% 301-399, level 3 noise
% 401-499, level 4 noise
opts.target_num = 1;
opts = loadDatasetInfo(opts);
if opts.dataset~=0
opts.save_path = "./paper_results/experiments/";
opts.path.mat_path = opts.path.matfiles_root + opts.path.event_name + "/";
end
checkDirectory(opts.save_path)
%% LiDAR properties
[LiDAR_opts, boundary] = loadLiDARProperties(opts);
%% target orientation
[object_list, ideal_object_list, target_properties] = loadTargetPose(opts, LiDAR_opts);
%% LiDAR Simulation
if opts.regen_LiDAR_points && opts.data.simulation
[object_list, LiDAR_opts, boundary] = genLiDARPointsGivenObjectList(opts, object_list, LiDAR_opts, boundary);
save(opts.save_path + target_properties.name + ".mat", 'object_list', 'LiDAR_opts', 'boundary', 'color_list')
disp("Generated LiDAR points on targets")
elseif ~opts.regen_LiDAR_points && opts.data.simulation
file_to_be_load = opts.load_path + target_properties.name + ".mat";
if isfile(file_to_be_load)
load(file_to_be_load)
disp("Loaded LiDAR points on targets")
else
fprintf("Can't find the mat file: %s \nGenerating a new one...\n", file_to_be_load)
[object_list, LiDAR_opts, boundary] = genLiDARPointsGivenObjectList(opts, object_list, LiDAR_opts, boundary);
save(opts.save_path + name + ".mat", 'object_list', 'LiDAR_opts', 'boundary', 'color_list')
disp("Generated LiDAR points on targets")
end
end
%%
opts.reuse_sampled = 1;
opts.max_iter = 15;
opts.downsample = 0;
opts.use_edge_points = 1;
opts.debug = 0;
opts.ncorrespondences.num_per_point = 1; % 1 ,2 ,4
opts.ncorrespondences.mode = 1; % 1: smallest dis 2: farthest dis
fig.save_figs = save_fig;
disp("user setting loaded")
%% tempalte
if opts.data.simulation && opts.use_best_shape
template = load("original_shape.mat");
rotatated_ideal = [300 0 0];
template.original_shape = ...
moveByRPYXYZ(template.original_shape, rotatated_ideal, [0 0 0]);
template.original_shape = template.original_shape(1:3, :);
centroid = mean(template.original_shape, 2);
offset = [0 0 0];
ideal_angle_list = [0, 0, 0];
scale = 1;
[ideal_object_list, ideal_color_list] = createOptimalShape(...
ideal_angle_list, -centroid', scale*template.original_shape);
elseif opts.data.simulation && ~opts.use_best_shape
%% ideal target pose
opts_obs.target_size = 1;
opts_obs.polygon = 4;
opts_obs.rpy = [0, 0, 0];
opts_obs.xyz = [0, 0, 0];
[ideal_object_list, ideal_color_list] = createDynamicScene(opts_obs);
end
disp("template loaded")
%% point cloud
if opts.data.simulation
pc_file = target_properties.name + ".mat";
original_pc = load(opts.load_path + pc_file);
org_num_points = size(original_pc.object_list.points_mat, 2);
else
original_pc.object_list = object_list;
original_pc.LiDAR_opts = LiDAR_opts;
org_num_points = size(original_pc.object_list.points_mat, 2);
end
if org_num_points < 10 && opts.data.simulation
file_to_delete = opts.load_path + pc_file;
delete(file_to_delete)
assert(org_num_points > 10, "the number of points are too less")
end
cur_pc = original_pc;
disp("LiDAR pc loaded")
%% calculate the edge points
if opts.data.simulation
edge_points = findEdgePointsFromLiDARPoints(original_pc);
else
edge_points = [];
for ring = 1: opts.data.num_rings
ring_points = original_pc.object_list.points_mat(:, original_pc.object_list.points_mat(5,:)==ring);
num_ring_points = size(ring_points, 2);
if num_ring_points > 0
edge_points = [edge_points, ring_points(1:3, 1)];
if num_ring_points > 1
edge_points = [edge_points, ring_points(1:3, end)];
end
end
end
end
% rpy = [0,0,0];
%
% xyz = [0,0,0];
% if opts.data.simulation
% [cur_pc.object_list, ~] = createOptimalShape(...
% rpy, xyz, convertXYZstructToXYZmatrix(...
% cur_pc.object_list.object_vertices));
% end
%
%
% H_trans = constructHByRPYXYZMovingPoints(rpy, xyz);
% cur_pc.object_list.points_mat = H_trans * ...
% convertToHomogeneousCoord(edge_points(1:3, :));
% initial_pc = cur_pc.object_list.points_mat;
% disp("pc transformed")
original_pc_points = original_pc.object_list.points_mat;
% if opts.data.simulation
% [original_pc.object_list, ~] = createOptimalShape(...
% rpy, xyz, convertXYZstructToXYZmatrix(...
% original_pc.object_list.object_vertices));
% end
% original_pc.object_list.points_mat = H_trans * ...
% convertToHomogeneousCoord(original_pc_points(1:3, :));
% disp("pc transformed")
original_plane_correspondences = createOptimalShapePlaneCorrespondences(...
original_pc.object_list.points_mat, ideal_object_list);
disp("formed original plane correspondences")
% gt
if opts.data.simulation
[d, ~, transform] = procrustes(...
original_pc.object_list.object_vertices_mat', ...
ideal_object_list.object_vertices_mat', ...
'scaling',false,'reflection',false);
assert(d<1e-5, "gt is not correct: %.4f", d)
original_pc.object_list.H = [transform.T' transform.c(1, :)'; 0 0 0 1];
gt_vertices = original_pc.object_list.H * ...
converToHomogeneousCoord(ideal_object_list.object_vertices_mat_h);
end
out_t = estimateLiDARPose(opts, cur_pc, ideal_object_list);
estimated_vertices = out_t.estimated_vertices;
solver = out_t.solver;
%%
disp("=================================")
disp("======= Estimated Results =======")
disp("=================================")
fprintf("Spent %.2f seconds\n", solver.computation_time)
if opts.data.simulation
RMSE = sqrt(norm(...
estimated_vertices - original_pc.object_list.object_vertices_mat_h, ...
'fro').^2/4);
RMSE_percentage = RMSE / norm(object_list.centroid) * 100;
gt_H_LT = original_pc.object_list.H
est_H_LT = inv(solver.H_final)
geo_dis = norm(Log_SE3(gt_H_LT / est_H_LT));
geo_rot = rad2deg(norm(Log_SO3(gt_H_LT(1:3, 1:3)/est_H_LT(1:3, 1:3))));
trans_error = norm(gt_H_LT(1:3, 4) - est_H_LT(1:3, 4));
trans_error_percentage = trans_error / norm(object_list.centroid) * 100;
% geo_dis = norm(Log_SE3(object_list.H_inv/solver.H_final));
% geo_rot = rad2deg(norm(Log_SO3(object_list.H_inv(1:3, 1:3)/solver.H_final(1:3, 1:3))));
% trans_error = norm(object_list.H_inv(1:3, 4) - solver.H_final(1:3, 4));
% trans_error_percentage = trans_error / norm(object_list.centroid) * 100;
fprintf("Geodesic Distance on SE(3) (norm(Log(H_gt * inv(H_est)))): %.4f \n", geo_dis)
fprintf("Geodesic Distance on SO(3) (norm(Log(R_gt * inv(R_est)))): %.4f [deg]\n", geo_rot)
fprintf("Translation Error: %.4f [m]\n", trans_error)
fprintf("Percentage of Translation Error: %.4f [%%] \n", trans_error_percentage)
fprintf("Percentage of Vertices Error: %.4f [%%] \n", RMSE_percentage)
fprintf("RMSE: %.2f [m]\n", RMSE)
else
fprintf("Target Distance: %.2f [m]\n", norm(object_list.centroid))
fprintf("original_cost: %.2f [m]\n", sum(abs(original_plane_correspondences.cost(Pose()))))
fprintf("optimized_cost: %.2f [m]\n", solver.optimized_cost)
end
%%
if opts.data.simulation
if ~isfile('results.csv')
fid = fopen('results.csv', 'A' );
fprintf(fid, 'x, y, z, r, p, h, rmse, rmse_per, geo_dis, geo_rot, trans_error, trans_error_per\n');
else
fid = fopen('results.csv', 'A' );
end
fprintf(fid, '%.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f, %.4f\n', ...
target_properties.translation_list(1), target_properties.translation_list(2), target_properties.translation_list(3), ...
target_properties.angle_list(1), target_properties.angle_list(2), target_properties.angle_list(3), ...
RMSE, RMSE_percentage, geo_dis, geo_rot, ...
trans_error, trans_error_percentage);
fclose( fid );
end
disp("=================================")
disp("ALL Done")