-
Notifications
You must be signed in to change notification settings - Fork 0
/
ngsim_manipulation_y.py
155 lines (141 loc) · 6.34 KB
/
ngsim_manipulation_y.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
'''
This is the version that take y as input instead of velocity.
contraction dataset to [0,1]
for using it, make sure that the name of file is ngsim_manipulation.py
'''
import numpy as np
import pandas as pd
from tqdm import tqdm
import os
rbm_timesteps = 10 # timesteps in every RBM visible layer
deg_superpose=5 # degree of superposition. if deg_superpose == rbm_timesteps, it means no superposition
class Data(object):
def __init__(self):
df = pd.read_csv("./ngsim_data/pretreatment-0750m-0805m.csv", sep=",")
self.data=df.loc[:, ['Local_X','Local_Y']].values
self.dataset=[] # a list of trajectories(normalized), every traj is a ndarray(time series)
i=0
width = (int)(self.data.shape[1]*rbm_timesteps)
while( i < len(self.data)):
idx=i+df.at[i,'Total_Frames']
if idx > len(self.data):
traj = self.data[i:,:]
traj = traj [ :(int)(np.floor(traj.shape[0]/rbm_timesteps)*rbm_timesteps) ]
j=0
traj_superposed=[]
while(j < (len(traj)-rbm_timesteps+1) ):
traj_superposed.append( np.reshape(traj[j:j+rbm_timesteps, :],[width]) )
j+=deg_superpose
self.dataset.append( np.array( traj_superposed) )
#traj = traj [ :(int)(np.floor(traj.shape[0]/rbm_timesteps)*rbm_timesteps) ]
#self.dataset.append( np.reshape(traj, [(int)(len(traj)/rbm_timesteps), width]) )
#self.dataset.append( self.data[i::5,:] )
else:
traj = self.data[i:idx,:]
traj = traj [ :(int)(np.floor(traj.shape[0]/rbm_timesteps)*rbm_timesteps) ]
j=0
traj_superposed=[]
while(j < (len(traj)-rbm_timesteps+1) ):
traj_superposed.append( np.reshape(traj[j:j+rbm_timesteps, :],[width]) )
j+=deg_superpose ##1 is the degree of superpose. if j += rbm_timesteps, it means no superpose
self.dataset.append( np.array(traj_superposed) )
#traj = traj [ :(int)(np.floor(traj.shape[0]/rbm_timesteps)*rbm_timesteps) ]
#self.dataset.append( np.reshape(traj, [(int)(len(traj)/rbm_timesteps), width]) )
#self.dataset.append( self.data[i:idx:5,:] )
#assert(df.at[i,'Vehicle_ID'] != df.at[idx,'Vehicle_ID'])
i = idx
self.currentposition=0
self.num_trajectories=len(self.dataset)
self.max=np.loadtxt("./ngsim_data/saved_max.csv", delimiter=',')
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!shape of one trajectoire is : ", self.dataset[0].shape)
def next_traj(self):
if self.currentposition == self.num_trajectories:
raise Exception('End of dataset')
traj = self.dataset[self.currentposition]
self.currentposition += 1
if self.currentposition == self.num_trajectories :
self.currentposition=0
return traj
def get_traj(self, numero):
if numero >= self.num_trajectories:
raise Exception("numero is bigger than total numbers")
traj=self.dataset[numero]
return traj
def get_trajs(self, numeros):
trajs=[]
for numero in numeros:
trajs.append(self.dataset[numero])
return trajs
def get_trajectories(self, start, end):
return self.dataset[start: end]
def decontraction(self, traj):
decontract = traj.copy()
decontract = decontract*np.tile(self.max , rbm_timesteps)
return decontract
'''
def denormalisation(self, traj):
denorm=traj.copy()
for j in range(len(std)):
if self.std[j]==0:
raise Exception("std[{}] is 0".format(j))
denorm=denorm*self.std
denorm += self.mean
return denorm
'''
class Dis_Data(object):
def __init__(self):
df = pd.read_csv("./ngsim_data/trajectories-0750am-0805am.csv", sep=",")
df_lane = df.loc[:, 'Lane_ID']
df_lane=pd.get_dummies(df_lane)
self.data=df_lane.values
np.savetxt("./output_folder/data.csv", self.data)
self.dataset=[] # a list of trajectories(normalized), every traj is a ndarray(time series)
i=0
while( i < len(self.data)):
idx=i+df.at[i,'Total_Frames']
if idx > len(self.data):
self.dataset.append(self.data[i::5,:])
else:
self.dataset.append(self.data[i:idx:5,:])
assert(df.at[i,'Vehicle_ID'] != df.at[idx,'Vehicle_ID'])
i = idx
self.currentposition=0
self.num_trajectories=len(self.dataset)
def next_traj(self):
if self.currentposition == self.num_trajectories:
raise Exception('End of dataset')
traj = self.dataset[self.currentposition]
self.currentposition += 1
if self.currentposition == self.num_trajectories :
self.currentposition=0
return traj
def get_traj(self, numero):
if numero >= self.num_trajectories:
raise Exception("numero is bigger than total numbers")
traj=self.dataset[numero]
return traj
def get_trajectories(self, num):
return self.dataset[:num]
def onehot_to_category(onehots):
catg = [np.argmax(onehot) for onehot in onehots]
return catg
def write_traj(traj):
if not os.path.isdir("./output_folder"):
os.makedirs("./output_folder")
df = pd.DataFrame(traj)
df.to_csv("./output_folder/reconstructed_trajectory.csv", sep=",", header=False)
def pre_treatment():
if not os.path.isdir("./ngsim_data"):
os.makedirs("./ngsim_data")
df = pd.read_csv("./ngsim_data/trajectories-0750am-0805am.csv", sep=",")
data=df.loc[:, ['Local_X','Local_Y']].values
def contraction(data):
max=np.amax(data, axis=0)
data=data/max
np.savetxt("./ngsim_data/saved_max.csv", max, delimiter=",")
return data
data=contraction(data)
df.loc[:, ['Local_X','Local_Y']] = data
df.to_csv("./ngsim_data/pretreatment-0750m-0805m.csv", sep=',', index=False)
if __name__=='__main__':
pre_treatment()