forked from movingpandas/movingpandas
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrajectory_sampler.py
167 lines (138 loc) · 7.73 KB
/
trajectory_sampler.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
# -*- coding: utf-8 -*-
"""
***************************************************************************
trajectory_sampler.py
---------------------
Date : December 2018
Copyright : (C) 2018 by Anita Graser
Email : [email protected]
***************************************************************************
* *
* This program is free software; you can redistribute it and/or modify *
* it under the terms of the GNU General Public License as published by *
* the Free Software Foundation; either version 2 of the License, or *
* (at your option) any later version. *
* *
***************************************************************************
"""
import random
import warnings
from datetime import timedelta
from shapely.geometry import Point
from geometry_utils import measure_distance_spherical
from trajectory import DIRECTION_COL_NAME, SPEED_COL_NAME
class TrajectorySample():
def __init__(self, id, start_timedelta, past_timedelta, future_timedelta, past_traj, future_pos, future_traj):
self.id = id
self.start_secs = start_timedelta.total_seconds()
self.past_secs = past_timedelta.total_seconds()
self.future_secs = future_timedelta.total_seconds()
self.past_traj = past_traj
self.future_pos = future_pos
self.future_traj = future_traj
def __str__(self):
return "{};{};{};{};{};{};{}".format(
self.id, self.start_secs, self.past_secs, self.future_secs, self.past_traj.to_linestring().wkt,
self.future_pos.wkt, self.future_traj.to_linestring().wkt)
class TrajectorySampler():
def __init__(self, traj, tolerance = timedelta(seconds=1)):
self.traj = traj
self.sample_counter = 0
self.tolerance = tolerance
def _is_sampling_possible(self, past_timedelta, future_timedelta, min_meters_per_sec = 0.3):
sample_duration = past_timedelta + future_timedelta
if self.traj.get_duration() < sample_duration:
warnings.warn("Trajectory {} is too short to extract {} seconds sample!".format(
self.traj.id, sample_duration.total_seconds()))
return False
self.traj.add_speed()
self.traj.df['next_ms'] = self.traj.df[SPEED_COL_NAME].shift(-1)
self.traj.df = self.traj.df[:-1]
above_speed_limit = self.traj.df[self.traj.df['next_ms'] > min_meters_per_sec]
if len(above_speed_limit) == 0:
warnings.warn("No data above specified speed limit!")
return False
return True
def _get_actually_available_times(self, start_time, past_time, future_time):
sample_times = []
for t in [start_time, past_time, future_time]:
#print("Testing {}".format(t))
row = self.traj.get_row_at(t)
if abs(row['t'] - t) > self.tolerance:
return False
else:
sample_times.append(row['t'])
return sample_times
def _get_sample_times(self, df, delta_t, first_move_time, past_timedelta, future_timedelta, randomize):
for t, row in df.iterrows():
if t > self.traj.get_end_time() - (past_timedelta + future_timedelta):
continue
if randomize:
if t < first_move_time + delta_t:
continue
delta_t += row['delta_t']
start_time = self.traj.get_row_at(first_move_time + past_timedelta + delta_t)['t']
start_timedelta = start_time - first_move_time
past_time = start_time - past_timedelta
future_time = start_time + future_timedelta
x = self._get_actually_available_times(start_time, past_time, future_time)
if x:
start_time, past_time, future_time = x
successful = True
return successful, start_time, past_time, future_time, start_timedelta
raise RuntimeError('Failed to get sample times! ')
def _get_time_of_first_move(self, min_meters_per_sec):
above_speed_limit = self.traj.df[self.traj.df['next_ms'] > min_meters_per_sec]
return above_speed_limit.index.min().to_pydatetime()
def _filter_df(self, first_move_time):
df = self.traj.df[self.traj.df.index >= first_move_time]
df.iat[0, df.columns.get_loc("delta_t")] = timedelta(seconds=0)
return df
def _match_sample_pattern_to_df(self, df, first_move_time, past_timedelta, future_timedelta, randomize):
secs = int((self.traj.get_duration() - future_timedelta).total_seconds())
if secs < 0:
raise RuntimeError("Failed to extract sample from trajectory {}!".format(self.traj.id))
if randomize:
number_of_retries = 3
random_start = random.randint(0, secs)
delta_t = timedelta(seconds=random_start)
else:
number_of_retries = 1
delta_t = timedelta(seconds=0)
successful = False
while not successful and number_of_retries > 0:
try:
successful, start_time, past_time, future_time, start_timedelta = self._get_sample_times(
df, delta_t, first_move_time, past_timedelta, future_timedelta, randomize)
except RuntimeError:
number_of_retries -= 1
random_start = random.randint(0, secs)
delta_t = timedelta(seconds=random_start)
if not successful:
raise RuntimeError("Failed to extract sample from trajectory {}!".format(self.traj.id))
return successful, start_time, past_time, future_time, start_timedelta
def _is_moving_sufficiently(self, traj, min_meters_per_sec, past_timedelta):
line_coords = traj.to_linestring().coords
covered_distance = measure_distance_spherical(Point(line_coords[0]), Point(line_coords[-1]))
if covered_distance < 0.5 * min_meters_per_sec * past_timedelta.total_seconds():
return False
else:
return True
def get_sample(self, past_timedelta, future_timedelta, min_meters_per_sec=0.3, randomize=False, future_traj_duration=timedelta(hours=0)):
if not self._is_sampling_possible(past_timedelta, future_timedelta, min_meters_per_sec):
raise RuntimeError("Cannot extract sample from this trajectory!")
first_move_time = self._get_time_of_first_move(min_meters_per_sec)
df = self._filter_df(first_move_time)
successful, start_time, past_time, future_time, start_timedelta = self._match_sample_pattern_to_df(
df, first_move_time, past_timedelta, future_timedelta, randomize)
future_pos = self.traj.get_position_at(future_time, method='nearest')
past_traj = self.traj.get_segment_between(past_time, start_time)
past_traj.context = self.traj.context
if self.traj.has_parent():
future_traj = self.traj.parent.get_segment_between(start_time, max(start_time+future_traj_duration, future_time))
else:
future_traj = self.traj.get_segment_between(start_time, max(start_time+future_traj_duration, future_time))
if not self._is_moving_sufficiently(past_traj, min_meters_per_sec, past_timedelta):
raise RuntimeError("Skipping sample {} since it there is not enough movement!".format(self.traj.id))
sample_id = "{}".format(self.traj.id)
return TrajectorySample(sample_id, start_timedelta, past_timedelta, future_timedelta, past_traj, future_pos, future_traj)