-
Notifications
You must be signed in to change notification settings - Fork 15
/
environment.py
173 lines (147 loc) · 6.35 KB
/
environment.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
from animate import Animator
from cars import Cars, TrafficLights
import navigation as nav
import numpy as np
import simulation as sim
# additional testing module
# import convergent_learner
class Env:
def __init__(self, n, graph, agent, dt, animate=False):
"""
initializes an environment for a car in the system
:param n: int: number of cars to simulate
:param graph: OGraph object from
:param agent: int: the ID of the car (agent)
:param animate: bool: if the environment is to be animated while learning
"""
self.N = n
self.num = None
self.graph = graph
self.fig, self.ax = self.graph.fig, self.graph.ax
self.agent = agent
self.dt = dt
self.animate = animate
self.animator = None
self.axis = self.ax.axis()
self.route_times = []
self.car_init_method = sim.init_random_node_start_location
# self.car_init_method = sim.init_culdesac_start_location
self.light_init_method = sim.init_traffic_lights
# self.car_init_method = convergent_learner.init_custom_agent
# self.light_init_method = convergent_learner.init_custom_lights
self.cars_object = Cars(self.car_init_method(self.N, self.graph), self.graph)
self.lights_object = TrafficLights(self.light_init_method(self.graph, prescale=40), self.graph)
self.high = 10
self.low = 2
self.shortest_route_thresh = 5
def reset(self, num):
"""
resets the environment
:param num: tuple: int, int
:return state: int
"""
# initialize cars every reset
init_cars = self.car_init_method(n=self.N, graph=self.graph)
self.cars_object = Cars(init_state=init_cars, graph=self.graph)
stateview = self.refresh_stateview()
state = stateview.determine_state()[0]
state = state.index(True)
if self.animate:
# init animator
self.num = num
self.animator = Animator(fig=self.fig, ax=self.ax, cars_object=self.cars_object,
lights_object=self.lights_object, num=self.num, focus=self.agent)
return state
def refresh_stateview(self):
"""
this function prepares a fresh depiction of what state the car is in
:return stateview: object
"""
stateview = nav.StateView(graph=self.graph, car_index=self.agent,
cars=self.cars_object.state, lights=self.lights_object.state)
return stateview
def initialize_custom_reset(self, alternate_route):
"""
resets the environment with a custom route for the agent
:param alternate_route: list: list of alternate route nodes for car agent
:return state: list: initial state of agent
"""
# initialize the car and light state objects
init_car_state = self.car_init_method(
graph=self.graph, n=self.N, car_id=self.agent, alternate_route=alternate_route
)
self.cars_object = Cars(init_state=init_car_state, graph=self.graph)
if self.animate:
# init animator
self.animator = Animator(fig=self.fig, ax=self.ax, cars_object=self.cars_object,
lights_object=self.lights_object, num=self.num)
stateview = self.refresh_stateview()
state = stateview.determine_state()[0]
state = state.index(True)
return state
def step(self, action, num):
"""
This function runs a full simulation of a car from origin to destination
(if action, then use the alternate route)
:param action: int: 0 or 1
:param num: tuple: the simulation number out of the total number of simulations
:return new_state, reward, done, _: list: the end of the return is free to contain debugging info
"""
debug_report = []
if self.animate:
self.animator.reset(self.num)
stateview = self.refresh_stateview()
state, new_route, new_xpath, new_ypath = stateview.determine_state()
if action:
new_state = self.initialize_custom_reset(alternate_route=(new_route, new_xpath, new_ypath))
else:
new_state = state.index(True)
# Run the simulation until the car reaches the destination
arrived = False
i = 0
while not arrived:
arrived = self.simulation_step(i)
i += 1
route_time = self.cars_object.state.loc[self.agent]['route-time']
self.route_times.append(route_time)
# TODO: need new way of identifying shortest route time.
if len(self.route_times) < self.shortest_route_thresh:
shortest_route_found_reward = 0
done = False
elif np.isclose(0, self.route_times[-1] - np.min(self.route_times), atol=5 * self.dt).all():
"""
If the route time achieved after the simulation is within 5 x dt second of the minimum time achieved.
Define this environment condition as having found the shortest route (locally).
"""
shortest_route_found_reward = self.high
done = True
else:
shortest_route_found_reward = 0
done = False
if num[0] < 1:
reward = 0
else:
time_delta = self.route_times[num[0] - 1] - self.route_times[num[0]] + shortest_route_found_reward
if time_delta > 0:
reward = time_delta
else:
reward = 0
return new_state, reward, done, debug_report
def simulation_step(self, i):
"""
make one step in the simulation and determine if the car has arrived at the destination
:param i: simulation step
:return arrived: bool
"""
frontview = nav.FrontView(self.cars_object.state.loc[self.agent], self.graph)
end_of_route = frontview.end_of_route()
if not end_of_route:
if self.animate:
self.animator.animate(i)
else:
self.lights_object.update(self.dt)
self.cars_object.update(self.dt, self.lights_object.state)
arrived = False
else:
arrived = True
return arrived