-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathsimulator.py
430 lines (380 loc) · 15 KB
/
simulator.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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
# ICRA 2019 Battlefield simulator.
# Based on Top-down car dynamics simulation from OpenAI Gym.
import math
import random
import sys
import Box2D
import gym
import numpy as np
import pyglet
from gym import spaces
from gym.utils import EzPickle, colorize, seeding
from pyglet import gl
from battlefield.body.obstacle import ICRALayout
from battlefield.body.robot import Robot
from battlefield.body.projectile import Projectile
from battlefield.referee.contact import ContactListener
from battlefield.referee.buff import AreaBuff
from battlefield.referee.supply import AreaSupply
from battlefield.sensor.capture import callback_capture
from utils import *
WINDOW_W = 1200
WINDOW_H = 1000
SCALE = 40.0 # Track scale
PLAYFIELD = 400/SCALE # Game over boundary
FPS = 30
ZOOM = 2.7 # Camera zoom
SCAN_RANGE = 5 # m
class ICRABattleField(gym.Env, EzPickle):
__pos_safe = [
[0.5, 0.5], [0.5, 2.0], [0.5, 3.0], [0.5, 4.5], # 0 1 2 3
[1.5, 0.5], [1.5, 3.0], [1.5, 4.5], # 4 5 6
[2.75, 0.5], [2.75, 2.0], [2.75, 3.0], [2.75, 4.5], # 7 8 9 10
[4.0, 1.75], [4.0, 3.25], # 11 12
[5.25, 0.5], [5.25, 2.0], [5.25, 3.0], [5.25, 4.5], # 13 14 15 16
[6.5, 0.5], [6.5, 2.0], [6.5, 4.5], # 17 18 19
[7.5, 0.5], [7.5, 2.0], [7.5, 3.0], [7.5, 4.5] # 20 21 22 23
]
__id_pos_linked = [
[1, 2, 3, 4], [0, 2, 3], [0, 1, 3, 5], [0, 1, 2, 6],
[0, 7], [2, 9], [3, 10],
[8, 9, 10, 4], [7, 9, 10, 11], [7, 8, 10, 5, 12], [7, 8, 9],
[8, 14], [9, 15],
[14, 15, 16, 17], [13, 15, 16, 18, 11, 11, 11, 11, 11], [
13, 14, 16, 12, 12, 12, 12, 12], [13, 14, 15, 19],
[13, 20], [14, 21], [16, 23],
[21, 22, 23, 17], [20, 22, 23, 18], [20, 21, 23], [20, 21, 22, 19]
]
def __init__(self):
EzPickle.__init__(self)
self.seed()
self.__contactListener_keepref = ContactListener(self)
self.__world = Box2D.b2World(
(0, 0), contactListener=self.__contactListener_keepref)
self.viewer = None
self.__robots = []
self.__robot_name = [ID_R1, ID_B1]
self.__obstacle = None
self.__area_buff = None
self.__projectile = None
self.__area_supply = None
self.__callback_autoaim = callback_capture()
self.reward = 0.0
self.prev_reward = 0.0
self.actions = None
self.state = None
def seed(self, seed=None):
self.np_random, seed = seeding.np_random(seed)
return [seed]
def _destroy(self):
for r in self.__robots:
if r:
r.destroy()
r = None
if self.__obstacle:
self.__obstacle.destroy()
self.__obstacle = None
if self.__projectile:
self.__projectile.destroy()
self.__projectile = None
def reset(self):
self._destroy()
self.reward = 0.0
self.prev_reward = 0.0
self.t = 0.0
random_index = random.randint(0, 23)
#random_index = 5
init_pos_0 = self.__pos_safe[random_index]
#init_pos_1 = self.__pos_safe[9]
init_pos_1 = self.__pos_safe[random.choice(
self.__id_pos_linked[random_index])]
print(init_pos_0, init_pos_1)
self.__R1 = Robot(self.__world, 0, init_pos_0, ID_R1)
self.__B1 = Robot(self.__world, 0, init_pos_1, ID_B1)
self.__robots = [self.__R1, self.__B1]
self.__obstacle = ICRALayout(self.__world)
self.__projectile = Projectile(self.__world)
self.__area_buff = AreaBuff()
self.__area_supply = AreaSupply()
self.state = [RobotState(init_pos_0), RobotState(init_pos_1)]
self.actions = [Action(), Action()]
self.reward = 0
return init_pos_1
# return self.step(None)[0]
def __step_contact(self):
contact_bullet_robot = self.__contactListener_keepref.collision_bullet_robot
contact_bullet_wall = self.__contactListener_keepref.collision_bullet_wall
contact_robot_wall = self.__contactListener_keepref.collision_robot_wall
contact_robot_robot = self.__contactListener_keepref.collision_robot_robot
for bullet, robot in contact_bullet_robot:
self.__projectile.destroyById(bullet.id)
if(self.__robots[robot.id].buff_left_time) > 0:
self.__robots[robot.id].lose_health(25)
else:
self.__robots[robot.id].lose_health(50)
for bullet in contact_bullet_wall:
self.__projectile.destroyById(bullet.id)
for robot in contact_robot_wall:
pass
# self.__robots[robot.id].lose_health(100)
for robot in contact_robot_robot:
self.__robots[robot.id].lose_health(10)
self.__contactListener_keepref.clean()
def _step_action(self, robot: Robot, action: Action):
# gas, rotate, transverse, rotate cloud terrance, shoot
robot.move_ahead_back(action.v_t)
robot.turn_left_right(action.angular)
robot.move_left_right(action.v_n)
if int(self.t * FPS) % (60 * FPS) == 0:
robot.refresh_supply_oppotunity()
if action.supply > 0.99:
action.supply = 0.0
if robot.if_supply_available():
robot.use_supply_oppotunity()
if self.__area_supply.if_in_area(robot):
robot.supply()
if action.shoot > 0.99 and int(self.t*FPS) % (FPS/5) == 1:
if(robot.if_left_projectile()):
angle, pos = robot.get_gun_angle_pos()
robot.shoot()
self.__projectile.shoot(angle, pos)
def _autoaim(self, robot: Robot, state: RobotState):
#detected = {}
scan_distance, scan_type = [], []
state.detect = False
for i in range(-135, 135, 2):
angle, pos = robot.get_angle_pos()
angle += i/180*math.pi
p1 = (pos[0] + 0.3*math.cos(angle), pos[1] + 0.3*math.sin(angle))
p2 = (pos[0] + SCAN_RANGE*math.cos(angle),
pos[1] + SCAN_RANGE*math.sin(angle))
self.__world.RayCast(self.__callback_autoaim, p1, p2)
scan_distance.append(self.__callback_autoaim.fraction)
u = self.__callback_autoaim.userData
if u is not None and u.type == "robot":
scan_type.append(1)
if -45 <= i <= 45:
if not state.detect:
robot.set_gimbal(angle)
state.detect = True
else:
scan_type.append(0)
state.scan = [scan_distance, scan_type]
def _update_robot_state(self, robot: Robot, state: RobotState):
state.pos = robot.get_pos()
state.health = robot.get_health()
state.angle = robot.get_angle()
state.velocity = robot.get_velocity()
state.angular = robot.get_angular()
def set_robot_action(self, robot_id, action: Action):
self.actions[robot_id] = action
def step(self, action: Action):
###### observe ######
for robot, state in zip(self.__robots, self.state):
self._autoaim(robot, state)
self._update_robot_state(robot, state)
###### action ######
self.set_robot_action(ID_R1, action)
for robot, action in zip(self.__robots, self.actions):
if action is not None:
self._step_action(robot, action)
robot.step(1.0/FPS)
self.__world.Step(1.0/FPS, 6*30, 2*30)
self.t += 1.0/FPS
###### Referee ######
self.__step_contact()
for robot in self.__robots:
self.__area_buff.detect(robot, self.t)
###### reward ######
step_reward = 0
done = False
# First step without action, called from reset()
if self.actions[ID_R1] is not None:
self.reward = (self.__robots[ID_R1].get_health() -
self.__robots[ID_B1].get_health()) / 4000.0
#self.reward += 10 * self.t * FPS
step_reward = self.reward - self.prev_reward
if self.state[ID_R1].detect:
step_reward += 1/3000
if self.__robots[ID_R1].get_health() <= 0:
done = True
#step_reward -= 1
if self.__robots[ID_B1].get_health() <= 0:
done = True
#step_reward += 1
#self.reward += step_reward
self.prev_reward = self.reward
return self.state, step_reward, done, {}
@staticmethod
def get_gl_text(x, y):
return pyglet.text.Label('0000', font_size=16, x=x, y=y,
anchor_x='left', anchor_y='center',
color=(255, 255, 255, 255))
def render(self, mode='god'):
if self.viewer is None:
from gym.envs.classic_control import rendering
self.viewer = rendering.Viewer(WINDOW_W, WINDOW_H)
self.time_label = self.get_gl_text(20, WINDOW_H * 5.0 / 40.0)
self.score_label = self.get_gl_text(520, WINDOW_H * 2.5 / 40.0)
self.health_label = self.get_gl_text(520, WINDOW_H * 3.5 / 40.0)
self.projectile_label = self.get_gl_text(
520, WINDOW_H * 4.5 / 40.0)
self.buff_left_time_label = self.get_gl_text(
520, WINDOW_H * 5.5 / 40.0)
self.buff_stay_time = self.get_gl_text(520, WINDOW_H * 6.5 / 40.0)
self.transform = rendering.Transform()
if "t" not in self.__dict__:
return # reset() not called yet
zoom = ZOOM*SCALE
scroll_x = 4.0
scroll_y = 0.0
angle = 0
self.transform.set_scale(zoom, zoom)
self.transform.set_translation(
WINDOW_W/2 - (scroll_x*zoom*math.cos(angle) -
scroll_y*zoom*math.sin(angle)),
WINDOW_H/4 - (scroll_x*zoom*math.sin(angle) + scroll_y*zoom*math.cos(angle)))
self.__obstacle.draw(self.viewer)
if mode == 'god':
for robot in self.__robots:
robot.draw(self.viewer)
elif mode == "fps":
self.__robots[ID_R1].draw(self.viewer)
self.__robots[ID_B1].draw(self.viewer)
self.__projectile.draw(self.viewer)
arr = None
win = self.viewer.window
if mode != 'state_pixels':
win.switch_to()
win.dispatch_events()
win.clear()
t = self.transform
gl.glViewport(0, 0, WINDOW_W, WINDOW_H)
t.enable()
self._render_background()
for geom in self.viewer.onetime_geoms:
geom.render()
t.disable()
self._render_indicators(WINDOW_W, WINDOW_H)
win.flip()
self.viewer.onetime_geoms = []
return arr
def close(self):
if self.viewer is not None:
self.viewer.close()
self.viewer = None
def _render_background(self):
gl.glBegin(gl.GL_QUADS)
gl.glColor4f(0.4, 0.8, 0.4, 1.0)
gl.glVertex3f(-PLAYFIELD, +PLAYFIELD, 0)
gl.glVertex3f(+PLAYFIELD, +PLAYFIELD, 0)
gl.glVertex3f(+PLAYFIELD, -PLAYFIELD, 0)
gl.glVertex3f(-PLAYFIELD, -PLAYFIELD, 0)
gl.glColor4f(0.4, 0.9, 0.4, 1.0)
k = PLAYFIELD/20.0
for x in range(-20, 20, 2):
for y in range(-20, 20, 2):
gl.glVertex3f(k*x + k, k*y + 0, 0)
gl.glVertex3f(k*x + 0, k*y + 0, 0)
gl.glVertex3f(k*x + 0, k*y + k, 0)
gl.glVertex3f(k*x + k, k*y + k, 0)
gl.glEnd()
self.__area_buff.render(gl)
self.__area_supply.render(gl)
def _render_indicators(self, W, H):
self.time_label.text = "Time: {} s".format(int(self.t))
self.score_label.text = "Score: %04i" % self.reward
self.health_label.text = "health left Car0 : {} Car1: {} ".format(
self.__robots[ID_R1].get_health(), self.__robots[ID_B1].get_health())
self.projectile_label.text = "Car0 bullets : {}, oppotunity to add : {} ".format(
self.__robots[ID_R1].get_left_projectile(
), self.__robots[ID_R1].supply_opportunity_left
)
self.buff_stay_time.text = 'Buff Stay Time: Red {}s, Blue {}s'.format(
int(self.__area_buff.get_single_buff(GROUP_RED).get_stay_time()),
int(self.__area_buff.get_single_buff(GROUP_BLUE).get_stay_time()))
self.buff_left_time_label.text = 'Buff Left Time: Red {}s, Blue {}s'.format(
int(self.__robots[ID_R1].buff_left_time),
int(self.__robots[ID_B1].buff_left_time))
self.time_label.draw()
self.score_label.draw()
self.health_label.draw()
self.projectile_label.draw()
self.buff_stay_time.draw()
self.buff_left_time_label.draw()
if __name__ == "__main__":
from pyglet.window import key, mouse
# gas, rotate, transverse, rotate cloud terrance, shoot, reload
#a = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0])
#target = [0, 0]
a = Action()
def on_mouse_release(x, y, button, modifiers):
x_low, x_high, y_low, y_high = 168, 1033, 249, 789
width = x_high - x_low
height = y_high - y_low
x = (x - x_low) / width * 8.0
y = (y - y_low) / height * 5.0
target[0] = x
target[1] = y
def key_press(k, mod):
global restart
if k == key.ESCAPE:
restart = True
if k == key.W:
a.v_t = +1.0
if k == key.S:
a.v_t = -1.0
if k == key.Q:
a.angular = +1.0
if k == key.E:
a.angular = -1.0
if k == key.D:
a.v_n = +1.0
if k == key.A:
a.v_n = -1.0
if k == key.SPACE:
a.shoot = +1.0
if k == key.R:
a.supply = +1.0
def key_release(k, mod):
if k == key.W:
a.v_t = +0.0
if k == key.S:
a.v_t = -0.0
if k == key.Q:
a.angular = +0.0
if k == key.E:
a.angular = -0.0
if k == key.D:
a.v_n = +0.0
if k == key.A:
a.v_n = -0.0
if k == key.SPACE:
a.shoot = +0.0
env = ICRABattleField()
env.render()
record_video = False
if record_video:
env.monitor.start('/tmp/video-test', force=True)
env.viewer.window.on_key_press = key_press
env.viewer.window.on_key_release = key_release
#env.viewer.window.on_mouse_release = on_mouse_release
#move = NaiveMove()
while True:
env.reset()
total_reward = 0.0
steps = 0
restart = False
s, r, done, info = env.step(a)
while True:
s, r, done, info = env.step(a)
total_reward += r
if steps % 200 == 0 or done:
print("step {} total_reward {:+0.2f}".format(steps, total_reward))
steps += 1
# Faster, but you can as well call env.render() every time to play full window.
if not record_video:
env.render()
if done or restart:
break
env.close()