forked from lawrie/fpga_roomba
-
Notifications
You must be signed in to change notification settings - Fork 0
/
OccupancyGrid.py
196 lines (183 loc) · 10.3 KB
/
OccupancyGrid.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
import json
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.cm as cm
class OccupancyGrid:
def __init__(self, mapXLength, mapYLength, initXY, unitGridSize, lidarFOV, numSamplesPerRev, lidarMaxRange, wallThickness):
xNum = int(mapXLength / unitGridSize)
yNum = int(mapYLength / unitGridSize)
x = np.linspace(-xNum * unitGridSize / 2, xNum * unitGridSize / 2, num=xNum + 1) + initXY['x']
y = np.linspace(-xNum * unitGridSize / 2, xNum * unitGridSize / 2, num=yNum + 1) + initXY['y']
self.OccupancyGridX, self.OccupancyGridY = np.meshgrid(x, y)
self.occupancyGridVisited = np.ones((xNum + 1, yNum + 1))
self.occupancyGridTotal = 2 * np.ones((xNum + 1, yNum + 1))
self.unitGridSize = unitGridSize
self.lidarFOV = lidarFOV
self.lidarMaxRange = lidarMaxRange
self.wallThickness = wallThickness
self.mapXLim = [self.OccupancyGridX[0, 0], self.OccupancyGridX[0, -1]]
self.mapYLim = [self.OccupancyGridY[0, 0], self.OccupancyGridY[-1, 0]]
self.numSamplesPerRev = numSamplesPerRev
self.angularStep = lidarFOV / numSamplesPerRev
self.numSpokes = int(np.rint(2 * np.pi / self.angularStep))
xGrid, yGrid, bearingIdxGrid, rangeIdxGrid = self.spokesGrid()
radByX, radByY, radByR = self.itemizeSpokesGrid(xGrid, yGrid, bearingIdxGrid, rangeIdxGrid)
self.radByX = radByX
self.radByY = radByY
self.radByR = radByR
# theta= 0 is x direction. spokes=0 is y direc tion, spokesStartIdx is the first ray of lidar scan direction. spokes increase counter-clockwise
self.spokesStartIdx = int(((self.numSpokes / 2 - self.numSamplesPerRev) / 2) % self.numSpokes)
def spokesGrid(self):
# 0th ray is at south, then counter-clock wise increases. Theta 0 is at east.
numHalfElem = int(self.lidarMaxRange / self.unitGridSize)
bearingIdxGrid = np.zeros((2 * numHalfElem + 1, 2 * numHalfElem + 1))
x = np.linspace(-self.lidarMaxRange, self.lidarMaxRange, 2 * numHalfElem + 1)
y = np.linspace(-self.lidarMaxRange, self.lidarMaxRange, 2 * numHalfElem + 1)
xGrid, yGrid = np.meshgrid(x, y)
bearingIdxGrid[:, numHalfElem + 1: 2 * numHalfElem + 1] = np.rint((np.pi / 2 + np.arctan(
yGrid[:, numHalfElem + 1: 2 * numHalfElem + 1] / xGrid[:, numHalfElem + 1: 2 * numHalfElem + 1]))
/ np.pi / 2 * self.numSpokes - 0.5).astype(int)
bearingIdxGrid[:, 0: numHalfElem] = np.fliplr(np.flipud(bearingIdxGrid))[:, 0: numHalfElem] + int(self.numSpokes / 2)
bearingIdxGrid[numHalfElem + 1: 2 * numHalfElem + 1, numHalfElem] = int(self.numSpokes / 2)
rangeIdxGrid = np.sqrt(xGrid**2 + yGrid**2)
return xGrid, yGrid, bearingIdxGrid, rangeIdxGrid
def itemizeSpokesGrid(self, xGrid, yGrid, bearingIdxGrid, rangeIdxGrid):
# Due to discretization, later theta added could lead to up to 1 deg discretization error
radByX = []
radByY = []
radByR = []
for i in range(self.numSpokes):
idx = np.argwhere(bearingIdxGrid == i)
radByX.append(xGrid[idx[:, 0], idx[:, 1]])
radByY.append(yGrid[idx[:, 0], idx[:, 1]])
radByR.append(rangeIdxGrid[idx[:, 0], idx[:, 1]])
return radByX, radByY, radByR
def expandOccupancyGridHelper(self, position, axis):
gridShape = self.occupancyGridVisited.shape
if axis == 0:
insertion = np.ones((int(gridShape[0] / 5), gridShape[1]))
if position == 0:
x = self.OccupancyGridX[0]
y = np.linspace(self.mapYLim[0] - int(gridShape[0] / 5) * self.unitGridSize, self.mapYLim[0],
num=int(gridShape[0] / 5), endpoint=False)
else:
x = self.OccupancyGridX[0]
y = np.linspace(self.mapYLim[1] + self.unitGridSize, self.mapYLim[1] + (int(gridShape[0] / 5) ) * self.unitGridSize,
num=int(gridShape[0] / 5), endpoint=False)
else:
insertion = np.ones((gridShape[0], int(gridShape[1] / 5)))
if position == 0:
y = self.OccupancyGridY[:, 0]
x = np.linspace(self.mapXLim[0] - int(gridShape[1] / 5) * self.unitGridSize, self.mapXLim[0],
num=int(gridShape[1] / 5), endpoint=False)
else:
y = self.OccupancyGridY[:, 0]
x = np.linspace(self.mapXLim[1] + self.unitGridSize, self.mapXLim[1] + (int(gridShape[1] / 5)) * self.unitGridSize,
num=int(gridShape[1] / 5), endpoint=False)
self.occupancyGridVisited = np.insert(self.occupancyGridVisited, [position], insertion, axis=axis)
self.occupancyGridTotal = np.insert(self.occupancyGridTotal, [position], 2 * insertion, axis=axis)
xv, yv = np.meshgrid(x, y)
self.OccupancyGridX = np.insert(self.OccupancyGridX, [position], xv, axis=axis)
self.OccupancyGridY = np.insert(self.OccupancyGridY, [position], yv, axis=axis)
self.mapXLim[0] = self.OccupancyGridX[0, 0]
self.mapXLim[1] = self.OccupancyGridX[0, -1]
self.mapYLim[0] = self.OccupancyGridY[0, 0]
self.mapYLim[1] = self.OccupancyGridY[-1, 0]
def expandOccupancyGrid(self, expandDirection):
gridShape = self.occupancyGridVisited.shape
if expandDirection == 1:
self.expandOccupancyGridHelper(0, 1)
elif expandDirection == 2:
self.expandOccupancyGridHelper(gridShape[1], 1)
elif expandDirection == 3:
self.expandOccupancyGridHelper(0, 0)
else:
self.expandOccupancyGridHelper(gridShape[0], 0)
def convertRealXYToMapIdx(self, x, y):
#mapXLim is (2,) array for left and right limit, same for mapYLim
xIdx = (np.rint((x - self.mapXLim[0]) / self.unitGridSize)).astype(int)
yIdx = (np.rint((y - self.mapYLim[0]) / self.unitGridSize)).astype(int)
return xIdx, yIdx
def checkMapToExpand(self, x, y):
if any(x < self.mapXLim[0]):
return 1
elif any(x > self.mapXLim[1]):
return 2
elif any(y < self.mapYLim[0]):
return 3
elif any(y > self.mapYLim[1]):
return 4
else:
return -1
def checkAndExapndOG(self, x, y):
"""check x, y (vector points) are inside OG. If not, expand OG."""
expandDirection = self.checkMapToExpand(x, y)
while (expandDirection != -1):
self.expandOccupancyGrid(expandDirection)
expandDirection = self.checkMapToExpand(x, y)
def updateOccupancyGrid(self, reading, dTheta = 0, update=True):
x, y, theta, rMeasure = reading['x'], reading['y'], reading['theta'], reading['range']
theta += dTheta
rMeasure = np.asarray(rMeasure)
spokesOffsetIdxByTheta = int(np.rint(theta / (2 * np.pi) * self.numSpokes))
emptyXList, emptyYList, occupiedXList, occupiedYList = [], [], [], []
for i in range(self.numSamplesPerRev):
spokeIdx = int(np.rint((self.spokesStartIdx + spokesOffsetIdxByTheta + i) % self.numSpokes))
xAtSpokeDir = self.radByX[spokeIdx]
yAtSpokeDir = self.radByY[spokeIdx]
rAtSpokeDir = self.radByR[spokeIdx]
if rMeasure[i] < self.lidarMaxRange:
emptyIdx = np.argwhere(rAtSpokeDir < rMeasure[i] - self.wallThickness / 2)
else:
emptyIdx = []
occupiedIdx = np.argwhere(
(rAtSpokeDir > rMeasure[i] - self.wallThickness / 2) & (rAtSpokeDir < rMeasure[i] + self.wallThickness / 2))
xEmptyIdx, yEmptyIdx = self.convertRealXYToMapIdx(x + xAtSpokeDir[emptyIdx], y + yAtSpokeDir[emptyIdx])
xOccupiedIdx, yOccupiedIdx = self.convertRealXYToMapIdx(x + xAtSpokeDir[occupiedIdx], y + yAtSpokeDir[occupiedIdx])
if update:
self.checkAndExapndOG(x + xAtSpokeDir[occupiedIdx], y + yAtSpokeDir[occupiedIdx])
if len(emptyIdx) != 0:
self.occupancyGridTotal[yEmptyIdx, xEmptyIdx] += 1
if len(occupiedIdx) != 0:
self.occupancyGridVisited[yOccupiedIdx, xOccupiedIdx] += 2
self.occupancyGridTotal[yOccupiedIdx, xOccupiedIdx] += 2
else:
emptyXList.extend(x + xAtSpokeDir[emptyIdx])
emptyYList.extend(y + yAtSpokeDir[emptyIdx])
occupiedXList.extend(x + xAtSpokeDir[occupiedIdx])
occupiedYList.extend(y + yAtSpokeDir[occupiedIdx])
if not update:
return np.asarray(emptyXList), np.asarray(emptyYList), np.asarray(occupiedXList), np.asarray(occupiedYList)
def plotOccupancyGrid(self, xRange = None, yRange= None, plotThreshold = True):
if xRange is None or xRange[0] < self.mapXLim[0] or xRange[1] > self.mapXLim[1]:
xRange = self.mapXLim
if yRange is None or yRange[0] < self.mapYLim[0] or yRange[1] > self.mapYLim[1]:
yRange = self.mapYLim
ogMap = self.occupancyGridVisited / self.occupancyGridTotal
xIdx, yIdx = self.convertRealXYToMapIdx(xRange, yRange)
ogMap = ogMap[yIdx[0]: yIdx[1], xIdx[0]: xIdx[1]]
ogMap = np.flipud(1 - ogMap)
plt.imshow(ogMap, cmap='gray', extent=[xRange[0], xRange[1], yRange[0], yRange[1]])
plt.show()
if plotThreshold:
ogMap = ogMap >= 0.5
plt.matshow(ogMap, cmap='gray', extent=[xRange[0], xRange[1], yRange[0], yRange[1]])
plt.show()
def updateTrajectoryPlot(matchedReading, xTrajectory, yTrajectory, colors, count):
x, y, theta, range = matchedReading['x'], matchedReading['y'], matchedReading['theta'], matchedReading['range']
xTrajectory.append(x)
yTrajectory.append(y)
if count % 1 == 0:
plt.scatter(x, y, color=next(colors), s=35)
def main():
initMapXLength, initMapYLength, unitGridSize, lidarFOV, lidarMaxRange = 10, 10, 0.02, np.pi, 10 # in Meters
wallThickness = 7 * unitGridSize
numSamplesPerRev = 450
initXY = {"x":0, "y":0, "theta":0}
og = OccupancyGrid(initMapXLength, initMapYLength, initXY, unitGridSize, lidarFOV, numSamplesPerRev, lidarMaxRange, wallThickness)
count = 0
plt.figure(figsize=(19.20, 19.20))
xTrajectory, yTrajectory = [], []
og.plotOccupancyGrid()
if __name__ == '__main__':
main()