-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathspaces.py
340 lines (254 loc) · 11.6 KB
/
spaces.py
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
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
import numpy as np
from PIL import Image
import time
from IPython.display import display, clear_output
import cv2
from mark_constants import *
from math_utils import *
#simple arm with 2 segments, note that theta refers to the angle as percentage of 2*pi (i.e. between 0. and 1.)
class Arm():
def __init__(self, theta1, theta2, l1, l2):
self.theta1, self.theta2, self.l1, self.l2 = theta1, theta2, l1, l2
def set_thetas(self, thetas):
self.theta1, self.theta2 = thetas[0], thetas[1]
def as_coordinates(self):
v1 = np.array([np.cos(self.theta1*np.pi*2), np.sin(self.theta1*np.pi*2)])
v2 = np.array([np.cos(self.theta2*np.pi*2), np.sin(self.theta2*np.pi*2)])
p1 = np.array([0.5, 0.2])
p2 = p1 + v1*self.l1
p3 = p2 + v2*self.l2
return p1, p2, p3
class Space():
def __init__(self, dims):
self.dims = tuple(dims)
self.dim = len(dims)
def in_bounds(self, p):
return (p >=0.0).all() and (p <= 1.0).all()
def in_bounds_index(self, index):
for k,i in enumerate(index):
if i<0 or i>=self.dims[k]: return False #TODO prettier with numpy ops?
return True
def at(self, point, as_array=False):
assert len(point) == self.dim
index = []
for i in range(len(self.dims)):
index.append(int(self.dims[i]*point[i]))
if index[i] == self.dims[i]: index[i] -= 1
if as_array: return np.array(index, dtype=np.int)
return tuple(index)
def draw_point(self, p, color):
pass
def draw_line(self, p1, p2, color):
pass
def display(self, path=None):
pass
def draw_path(self, path, color_path=LIGHT_GREEN, color_start=BLUE, color_goal=LIGHT_BLUE):
for i in range(len(path)-1):
self.draw_line(path[i], path[i+1], color_path)
self.draw_point(path[0], color_start)
self.draw_point(path[-1], color_goal)
class GridSpace(Space):
def __init__(self, dims):
Space.__init__(self, dims)
dims = list(dims)
self.grid = np.zeros(dims, dtype=np.uint8)
dims.append(3)
self.grid_vis = np.zeros(dims, dtype=np.uint8) +255
self.checkpoint_vis()
def checkpoint_vis(self):
self._checkpoint_vis = np.copy(self.grid_vis)
def reset_vis(self):
self.grid_vis = np.copy(self._checkpoint_vis)
#copy the current grid as grey image to grid_vis, used by voronoi diagram
def grid_to_vis(self):
self.grid_vis = np.stack((self.grid,)*3, -1)
def draw_point(self, p, color):
self.mark(p, mark=None, mark_vis=color)
def draw_line(self, p1, p2, color):
self.add_line(p1, p2, mark=None, mark_vis=color)
def mark(self, point, mark=None, mark_vis=None):
if not self.in_bounds(np.array(point)): return
if mark is not None: self.grid[self.at(point)] = mark
if mark_vis is not None: self.grid_vis[self.at(point)] = mark_vis
def mark_index(self, index, mark=None, mark_vis=None):
if not self.in_bounds_index(index): return
if mark is not None: self.grid[tuple(index)] = mark
if mark_vis is not None: self.grid_vis[tuple(index)] = mark_vis
def occupied(self, point=None):
if not self.in_bounds(point): return True
return self.grid[self.at(point)] == OCCUPIED
def occupied_index(self, index):
if not self.in_bounds_index(index): return True
return self.grid[tuple(index)] == OCCUPIED
def check_arm(self, arm):
p1, p2, p3 = arm.as_coordinates()
return self.check_line(p1, p2) and self.check_line(p2, p3)
def check_line(self, p1, p2):
assert self.dim == 2
if not(self.in_bounds(p1) and self.in_bounds(p2)): return False
p1, p2 = np.array(p1), np.array(p2)
#smallest step size: size of a cell dx,dy
dx,dy = 1.0/self.dims[0],1.0/self.dims[1]
d = min(dx,dy)
v = p2-p1
v_max = np.max(np.abs(v))
steps = int(np.ceil(v_max/d))
v = v / steps
for i in range(steps+1):
if self.occupied(p1 + i*v): return False
return True
def add_arm(self, arm, mark=ARM, mark_vis=ARM_VIS):
p1, p2, p3 = arm.as_coordinates()
self.add_line(p1, p2, mark=mark, mark_vis=mark_vis)
self.add_line(p2, p3, mark=mark, mark_vis=mark_vis)
def add_line(self, p1, p2, mark=None, mark_vis=None):
assert self.dim == 2
p1,p2 = np.array(p1),np.array(p2)
#smallest step size: size of a cell dx,dy
dx,dy = 1.0/self.dims[0],1.0/self.dims[1]
d = min(dx,dy)
v = p2-p1
v_max = np.max(np.abs(v))
steps = int(np.ceil(v_max/d))+1
v = v / steps
for i in range(steps):
p = p1 + i*v
if not self.in_bounds(p): return
self.mark(p, mark, mark_vis)
def add_rect(self, rect, mark=OCCUPIED, mark_vis=OCCUPIED_VIS):
assert self.dim == 2
start = self.at([rect.x, rect.y])
end = self.at([rect.x+rect.width, rect.y+rect.height])
self.grid[start[0]:end[0], start[1]:end[1]] = mark
self.grid_vis[start[0]:end[0], start[1]:end[1]] = mark_vis
def animate_arm(self, arm, path, video_file=None, secs_at_start=1, secs_at_end=1):
if video_file is not None:
fourcc = cv2.VideoWriter_fourcc(*'MPEG')
writer = cv2.VideoWriter(video_file, fourcc, 30.0, (500,500))
#always take secs_per_point seconds between two points in the path (in the animation)
secs_per_point = 1
hertz = 30
frames = hertz * secs_per_point
dt = 1./frames
if video_file is not None:
arm.set_thetas(path[0])
self.add_arm(arm, mark=None, mark_vis=BLUE)
arm.set_thetas(path[-1])
self.add_arm(arm, mark=None, mark_vis=LIGHT_BLUE)
img = np.array(self.display())
img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
for i in range(secs_at_start*hertz):
writer.write(img)
for i in range(len(path)-1):
p1, p2 = path[i], path[i+1]
v = p2 - p1
for frame in range(frames):
#draw arm
self.reset_vis()
p = p1 + v*frame*dt
arm.set_thetas(p)
self.add_arm(arm, mark=None)
#display/write to video
img = np.array(self.display())
if video_file is not None:
img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
writer.write(img)
time.sleep(dt)
if video_file is not None:
for i in range(secs_at_end*hertz):
writer.write(img)
writer.release()
def display(self, path=None, size=(500,500)):
img = np.swapaxes(self.grid_vis, 1, 0)
img = Image.fromarray(img, 'RGB')
img = img.resize(size)
clear_output(wait=True)
display(img)
if path is not None: img.save(path)
return img
#note: implementation assumes the edges of the polygon are free space
class PolygonSpace(Space):
def __init__(self, dim):
Space.__init__(self, [100]*dim)
self.polygons = []
self.points_to_draw = []
self.lines_to_draw = []
def draw_point(self, p, color):
self.points_to_draw.append((p, color))
def draw_line(self, p1, p2, color):
self.lines_to_draw.append((p1,p2,color))
def add_polygon(self, points):
#calculate line segments in homogenous coords
polylines = []
for i in range(len(points)-1):
p1, p2 = to_hom(points[i]), to_hom(points[i+1])
polylines.append(to_line(p1,p2))
self.polygons.append((points, polylines))
#NOTE: if you have a circle as a robot and you're trying to check if it collides with an object when going along a line,
#then instead you could model the object bigger (extend it by the radius of the robot) and the robot only as a point
#the normal check_line here would then work
def check_line(self, p1, p2):
assert self.dim == 2
line = to_line(to_hom(p1), to_hom(p2))
#check for collisions with all polygons
for points,polylines in self.polygons:
for i in range(len(points)-1):
crosspoint = calculate_crosspoint(line, polylines[i], p1, p2, points[i], points[i+1])
if crosspoint is None: continue
#get distance to closest point to check if crosspoint actually is one of those points (happens when you're exactly at one of the polygons)
min_distance = 1 #only interested if mindistance < 1e-09, so this is fine
for p in [p1, p2, points[i], points[i+1]]: min_distance = min(min_distance, simple_distance(p, crosspoint))
if min_distance > 1e-09: return False
return True
def find_crosspoints(self, p1, p2):
assert self.dim == 2
p1, p2 = to_hom(p1), to_hom(p2)
line = to_line(p1,p2)
crosspoints = []
for points,polylines in self.polygons:
for i in range(len(points)-1):
crosspoint = calculate_crosspoint(line, polylines[i], p1, p2, points[i], points[i+1])
if crosspoint is not None:
crosspoints.append((crosspoint, i, (points,polylines)))
return crosspoints
def check_circle(self, p, r):
p = np.array(p)
for points,_ in self.polygons:
for polyp in points:
if np.linalg.norm(polyp - p) < r: return False
return True
def display(self, path=None, empty_lists=False):
#for now abuse grid space for this...
space = GridSpace([250]*self.dim)
for points,polylines in self.polygons:
for i in range(len(points)-1):
space.draw_line(points[i], points[i+1], color=OCCUPIED_VIS)
for line in self.lines_to_draw:
space.draw_line(line[0], line[1], color=line[2])
for point in self.points_to_draw:
space.draw_point(point[0], color=point[1])
if empty_lists:
self.lines_to_draw = []
self.points_to_draw = []
return space.display(path)
def to_grid_space(self, dims):
assert len(dims) == self.dim
space = GridSpace(dims)
for points,_ in self.polygons:
for i in range(len(points)-1):
space.add_line(points[i], points[i+1], mark=OCCUPIED, mark_vis=OCCUPIED_VIS)
return space
class Rect():
def __init__(self, x, y, width, height):
self.x, self.y, self.width, self.height = x, y, width, height
#create c-space given the space and arm, dims refers to size of the c-space
def create_arm_c_space(space, arm, dims=[500,500]):
assert len(dims) == 2 and space.dim == 2
c_space = GridSpace(dims)
for t1 in range(dims[0]):
for t2 in range(dims[1]):
arm.set_thetas((float(t1)/dims[0], float(t2)/dims[1]))
free = space.check_arm(arm)
if not free:
c_space.mark_index((t1,t2), mark=OCCUPIED, mark_vis=OCCUPIED_VIS)
return c_space