-
Notifications
You must be signed in to change notification settings - Fork 1
/
main.py
277 lines (215 loc) · 9.56 KB
/
main.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
__author__ = "Yska"
# Importing the required libraries
import sys
import argparse
import numpy as np
import math
from utils import move
import random
import matplotlib.pyplot as plt
from QuadMDP.QuadTree import Point, Rect
from QuadMDP.QuadMDP import QuadMDP
from utils.Obstacle import Obstacle, Agent
from utils.Global import Global
from utils.agent_graph import *
from matplotlib.animation import FuncAnimation
import time
import csv
def path(args):
# Initializing the world from the picture
filename = args.input
image = plt.imread('map/' + filename)
mapWidth, mapHeight = image.shape
print("Map Width:", mapWidth)
print("Map height:", mapHeight)
mapDepth = (int)(math.log(mapWidth,2))
searchDepth = 2 # Search depth within QuadMDP
# Global class for the whole environment
g = Global((mapWidth,mapHeight))
# QuadMDP class for generating the quad decomposed states
quad = QuadMDP(Rect(mapWidth/2-0.5,mapHeight/2-0.5,mapWidth,mapHeight),mapDepth)
# Initialize time collection
times = [0] #first entry is time steps, second+third are realtime global path search
# Load all obstacles
for x in range(mapWidth):
for y in range(mapHeight):
if image[y,x] == 0:
g.createObstacle(move.NONE,(x,y))
quad.insert(Point(x,y,True))
# Generate quad decomposed states
S = quad.findEmptySpace(searchDepth)
graph = quad.generateGraph(S, searchDepth)
# Generate the start position and the goal
startPos = (8,8)
goalPos = (119,8)
startQuadMDP = quad.findContainedQuadMDP(startPos)
goalQuadMDP = quad.findContainedQuadMDP(goalPos)
DFS = False # set to True to use DFS instead of BFS
Simple = True
# Get the optimal path in the Quadtree representation
if DFS: # DFS
path_dfs = quad.getPathDFSV2(S,searchDepth,startQuadMDP,goalQuadMDP, path = [], visited = set())
path = [element.getTuple() for element in path_dfs]
else: # BFS
times.append(time.perf_counter())
path = quad.getOptimalPath(S,searchDepth,startQuadMDP.getTuple(),goalQuadMDP.getTuple())
times.append(time.perf_counter())
path.append(goalPos)
# Projecting the path to the Global representation
for i, p in enumerate(path):
decision = np.random.randint(0, 3)
if decision == 0:
path[i] = (math.floor(p[0]), math.floor(p[1]))
if decision == 1:
path[i] = (math.ceil(p[0]), math.floor(p[1]))
if decision == 2:
path[i] = (math.floor(p[0]), math.ceil(p[1]))
if decision == 3:
path[i] = (math.ceil(p[0]), math.ceil(p[1]))
print("Path we want to take: ", path)
# Create the moving obstacles
moving_obstacles = [] # store the moving obstacles
moving_obstacles_positions = dict() # store the positions of the moving obstacles over time
num_obstacles = 200
a = np.random.randint(0, [[mapWidth],[mapHeight]], size = (2,num_obstacles)).T
for element in a: # creating random moving obstacles
obs1 = g.createObstacle(move.RANDOM,tuple(element)) # moving obstacles with random dynamics
moving_obstacles.append(obs1)
moving_obstacles_positions[obs1] = [tuple(element)]
# Create the agent
agent = g.createAgent(startPos)
agent.range = 2
# Print the initial state of the world
print("Initial position")
g.plot()
# Retrieve the dimensions of the picture and the first observation
Xlim, Ylim = mapWidth, mapHeight
obs = g.observe()
# Build observation environment
Graph = agent_graph(agent, obs, Xlim, Ylim)
# Initialization
idx_goal = 0 # index of the intermediary goal we are considering
complete_path = []
current_state = tuple(agent.location)
not_moving_count = 0
while idx_goal < len(path):
# Grab the current goal and project it to the observation environment
goal = tuple(path[idx_goal])
projected_goal = Graph.project_to_surroundings(goal)
# Get the optimal path on the low level
if DFS: # DFS
inter_path = Graph.get_path_DFS(current_state, projected_goal, [], set())
else: # BFS
inter_path = Graph.get_optimal_path(current_state, projected_goal)
# Get to the next state on the path
next_state = inter_path[1]
action = np.array(next_state) - np.array(current_state)
# Check that we are not stuck in a corner
if action == move.NONE:
not_moving_count += 1
else:
not_moving_count = 0
# If stuck for some reason, go back to the previous goal
if not_moving_count > 10 and action == move.NONE and idx_goal >= 1:
idx_goal -= 1
not_moving_count = 0
# Updating the state of the agents and the moving obstacles
agent.action = action
g.next()
times[0] += 1
# Position of the agent after moving
current_state = tuple(agent.location)
complete_path += [current_state]
print(current_state)
print(goal)
# Check how the environment look now
#g.plot() # deactivate for gain of time
# Get new observation
obs = g.observe()
print(obs)
# Tracking the positions of the obstacles
for obstacle in moving_obstacles:
moving_obstacles_positions[obstacle].append(tuple(obstacle.location))
# Load the new graph corresponding to the new positions of the agent and its new observation
Graph = agent_graph(agent, obs, Xlim, Ylim)
if current_state == goal: # If we have reached the goal, we go to the next one
idx_goal += 1
# Upload the entire trajectory so as to plot it
g.complete_path = complete_path
# Initialize the picture for plotting
figsize = 4
fig, ax = plt.subplots(1, 1, figsize=(g.x_dim/g.y_dim*figsize, figsize))
plt.title("time = " + str(g.time))
ax.set_xlim([-0.5,g.x_dim-0.5])
ax.set_ylim([-0.5,g.y_dim-0.5])
ax.set_aspect('equal')
ax.invert_yaxis()
plt.tight_layout()
# Plot the fixed obstacles outside of plot_agent so that they remain in the figure the entire simulation
for o in g.obstacles:
if moving_obstacles_positions.get(o) is None:
x,y = o.location
rx = [x-0.5,x+0.5,x+0.5,x-0.5]
ry = [y-0.5,y-0.5,y+0.5,y+0.5]
plt.fill(rx,ry,color='red',zorder=-2, alpha=0.8)
# plotting the global path to follow in blue outside of plot_agent so that they remain in the figure the entire simulation
for state in path:
x, y = state
rx = [x-0.5,x+0.5,x+0.5,x-0.5]
ry = [y-0.5,y-0.5,y+0.5,y+0.5]
plt.fill(rx,ry,color='blue',zorder=-2, alpha=1)
# Keeping track of specific artist elements being plotted so as to replace them at each iteration of plot_agent
moving_obstacles_artists_list = [] # made to contain the plotted positions of the moving obstacles
observation_artist_list = [] # made to contain the plotted observations
def plot_agent(i): # i = frame number
# iterative plotting function
plt.title("time = " + str(i))
for element in moving_obstacles_artists_list: # remove from the figure the previous positions of the moving obstacles
if len(element) > 0:
line = element.pop(0) # remove from the list
line.remove() # remove from the figure
# Plotting the agents position at time i
a = g.agents[0]
x,y = g.complete_path[i]
rx = [x-0.5,x+0.5,x+0.5,x-0.5]
ry = [y-0.5,y-0.5,y+0.5,y+0.5]
plt.fill(rx,ry,color='skyblue',zorder=-2, alpha=0.8)
# Plotting the observation at time i
ax_min = max(0, x-a.range)
ax_max = min(g.x_dim-1, x+a.range)
ay_min = max(0, y-a.range)
ay_max = min(g.y_dim-1, y+a.range)
rx = [ax_min-0.5, ax_max+0.5, ax_max+0.5, ax_min-0.5]
ry = [ay_min-0.5, ay_min-0.5, ay_max+0.5, ay_max+0.5]
for element in observation_artist_list:
if len(element)> 0: # remove the previous observation from the plot
line = element.pop(0)
line.remove()
observation_artist_list.append(plt.fill(rx, ry, color='lightgreen', zorder=-2, alpha=0.4)) # add the new observation to the list of plotted elements (to remove it next iteration)
# plotting moving obstacles
for o in moving_obstacles:
x,y = moving_obstacles_positions[o][i]
rx = [x-0.5,x+0.5,x+0.5,x-0.5]
ry = [y-0.5,y-0.5,y+0.5,y+0.5]
ln = plt.fill(rx,ry,color='red',zorder=-2, alpha=0.8) # plotting
moving_obstacles_artists_list.append(ln) # adding the artist so we can remove it at the next iteration
# building the animation object
ani = FuncAnimation(fig, plot_agent, frames=len(complete_path), interval=100, repeat=False)
# Either show or save as a gif (one cancels the other so you need to execute only one)
plt.show()
#ani.save('animation%4f.gif' % (time.time()), writer='imagemagick', fps=10)
return times
def main(args):
runs = []
with open('qtree.csv', 'w') as f:
writer = csv.writer(f)
for i in range(25):
runs.append(path(args))
writer.writerow(path(args))
print(runs)
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Quad Decomposition Demo')
parser.add_argument('-i', '--input', help='Grayscale PNG map file name', default='1.png')
args = parser.parse_args()
#main(args)
path(args)