-
Notifications
You must be signed in to change notification settings - Fork 384
/
FlockAviary.py
250 lines (214 loc) · 11.1 KB
/
FlockAviary.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
import math
import numpy as np
from gym_pybullet_drones.utils.enums import DroneModel, Physics
from gym_pybullet_drones.envs.single_agent_rl.BaseSingleAgentAviary import ActionType, ObservationType
from gym_pybullet_drones.envs.multi_agent_rl.BaseMultiagentAviary import BaseMultiagentAviary
class FlockAviary(BaseMultiagentAviary):
"""Multi-agent RL problem: flocking."""
################################################################################
def __init__(self,
drone_model: DroneModel=DroneModel.CF2X,
num_drones: int=2,
neighbourhood_radius: float=np.inf,
initial_xyzs=None,
initial_rpys=None,
physics: Physics=Physics.PYB,
freq: int=240,
aggregate_phy_steps: int=1,
gui=False,
record=False,
obs: ObservationType=ObservationType.KIN,
act: ActionType=ActionType.RPM):
"""Initialization of a multi-agent RL environment.
Using the generic multi-agent RL superclass.
Parameters
----------
drone_model : DroneModel, optional
The desired drone type (detailed in an .urdf file in folder `assets`).
num_drones : int, optional
The desired number of drones in the aviary.
neighbourhood_radius : float, optional
Radius used to compute the drones' adjacency matrix, in meters.
initial_xyzs: ndarray | None, optional
(NUM_DRONES, 3)-shaped array containing the initial XYZ position of the drones.
initial_rpys: ndarray | None, optional
(NUM_DRONES, 3)-shaped array containing the initial orientations of the drones (in radians).
physics : Physics, optional
The desired implementation of PyBullet physics/custom dynamics.
freq : int, optional
The frequency (Hz) at which the physics engine steps.
aggregate_phy_steps : int, optional
The number of physics steps within one call to `BaseAviary.step()`.
gui : bool, optional
Whether to use PyBullet's GUI.
record : bool, optional
Whether to save a video of the simulation in folder `files/videos/`.
obs : ObservationType, optional
The type of observation space (kinematic information or vision)
act : ActionType, optional
The type of action space (1 or 3D; RPMS, thurst and torques, or waypoint with PID control)
"""
super().__init__(drone_model=drone_model,
num_drones=num_drones,
neighbourhood_radius=neighbourhood_radius,
initial_xyzs=initial_xyzs,
initial_rpys=initial_rpys,
physics=physics,
freq=freq,
aggregate_phy_steps=aggregate_phy_steps,
gui=gui,
record=record,
obs=obs,
act=act
)
################################################################################
def _computeReward(self):
"""Computes the current reward value(s).
Returns
-------
dict[int, float]
The reward value for each drone.
"""
rewards = {}
states = np.array([self._getDroneStateVector(i) for i in range(self.NUM_DRONES)])
rewards[0] = -1 * np.linalg.norm(np.array([0, 0, 1]) - states[0, 0:3])**2
for i in range(1, self.NUM_DRONES):
rewards[i] = -1 * np.linalg.norm(states[i-1, 2] - states[i, 2])**2
return rewards
"""
# obs here is dictionary of the form {"i":{"state": Box(20,), "neighbors": MultiBinary(NUM_DRONES)}}
# parse velocity and position
vel = np.zeros((1, self.NUM_DRONES, 3)); pos = np.zeros((1, self.NUM_DRONES, 3))
for i in range(self.NUM_DRONES):
pos[0,i,:] = obs[ i ]["state"][0:3]
vel[0,i,:] = obs[ i ]["state"][10:13]
# compute metrics
# velocity alignment
ali = 0
EPSILON = 1e-3 # avoid divide by zero
linear_vel_norm = np.linalg.norm(vel, axis=2)
for i in range(self.NUM_DRONES):
for j in range(self.NUM_DRONES):
if j != i:
d = np.einsum('ij,ij->i', vel[:, i, :], vel[:, j, :])
ali += (d / (linear_vel_norm[:, i] + EPSILON) / (linear_vel_norm[:, j] + EPSILON))
ali /= (self.NUM_DRONES * (self.NUM_DRONES - 1))
# flocking speed
cof_v = np.mean(vel, axis=1) # center of flock speed
avg_flock_linear_speed = np.linalg.norm(cof_v, axis=-1)
# spacing
whole_flock_spacing = []
for i in range(self.NUM_DRONES):
flck_neighbor_pos = np.delete(pos, [i], 1)
drone_neighbor_pos_diff = flck_neighbor_pos - np.reshape(pos[:, i, :], (pos[:, i, :].shape[0], 1, -1))
drone_neighbor_dis = np.linalg.norm(drone_neighbor_pos_diff, axis=-1)
drone_spacing = np.amin(drone_neighbor_dis, axis=-1)
whole_flock_spacing.append(drone_spacing)
whole_flock_spacing = np.stack(whole_flock_spacing, axis=-1)
avg_flock_spacing = np.mean(whole_flock_spacing, axis=-1)
var_flock_spacing = np.var(whole_flock_spacing, axis=-1)
# flocking metrics
FLOCK_SPACING_MIN = 1.0; FLOCK_SPACING_MAX = 3.0
if FLOCK_SPACING_MIN < avg_flock_spacing[0] < FLOCK_SPACING_MAX:
avg_flock_spac_rew = 0.0
else:
avg_flock_spac_rew = min(math.fabs(avg_flock_spacing[0] - FLOCK_SPACING_MIN),
math.fabs(avg_flock_spacing[0] - FLOCK_SPACING_MAX))
reward = ali[0] + avg_flock_linear_speed[0] - avg_flock_spac_rew - var_flock_spacing[0]
return { i : reward for i in range(self.NUM_DRONES) }
"""
################################################################################
def _computeDone(self):
"""Computes the current done value(s).
Returns
-------
dict[int | "__all__", bool]
Dictionary with the done value of each drone and
one additional boolean value for key "__all__".
"""
bool_val = True if self.step_counter/self.SIM_FREQ > self.EPISODE_LEN_SEC else False
done = {i: bool_val for i in range(self.NUM_DRONES)}
done["__all__"] = True if True in done.values() else False
return done
################################################################################
def _computeInfo(self):
"""Computes the current info dict(s).
Unused.
Returns
-------
dict[int, dict[]]
Dictionary of empty dictionaries.
"""
return {i: {} for i in range(self.NUM_DRONES)}
################################################################################
def _clipAndNormalizeState(self,
state
):
"""Normalizes a drone's state to the [-1,1] range.
Parameters
----------
state : ndarray
(20,)-shaped array of floats containing the non-normalized state of a single drone.
Returns
-------
ndarray
(20,)-shaped array of floats containing the normalized state of a single drone.
"""
MAX_LIN_VEL_XY = 3
MAX_LIN_VEL_Z = 1
MAX_XY = MAX_LIN_VEL_XY*self.EPISODE_LEN_SEC
MAX_Z = MAX_LIN_VEL_Z*self.EPISODE_LEN_SEC
MAX_PITCH_ROLL = np.pi # Full range
clipped_pos_xy = np.clip(state[0:2], -MAX_XY, MAX_XY)
clipped_pos_z = np.clip(state[2], 0, MAX_Z)
clipped_rp = np.clip(state[7:9], -MAX_PITCH_ROLL, MAX_PITCH_ROLL)
clipped_vel_xy = np.clip(state[10:12], -MAX_LIN_VEL_XY, MAX_LIN_VEL_XY)
clipped_vel_z = np.clip(state[12], -MAX_LIN_VEL_Z, MAX_LIN_VEL_Z)
if self.GUI:
self._clipAndNormalizeStateWarning(state,
clipped_pos_xy,
clipped_pos_z,
clipped_rp,
clipped_vel_xy,
clipped_vel_z
)
normalized_pos_xy = clipped_pos_xy / MAX_XY
normalized_pos_z = clipped_pos_z / MAX_Z
normalized_rp = clipped_rp / MAX_PITCH_ROLL
normalized_y = state[9] / np.pi # No reason to clip
normalized_vel_xy = clipped_vel_xy / MAX_LIN_VEL_XY
normalized_vel_z = clipped_vel_z / MAX_LIN_VEL_XY
normalized_ang_vel = state[13:16]/np.linalg.norm(state[13:16]) if np.linalg.norm(state[13:16]) != 0 else state[13:16]
norm_and_clipped = np.hstack([normalized_pos_xy,
normalized_pos_z,
state[3:7],
normalized_rp,
normalized_y,
normalized_vel_xy,
normalized_vel_z,
normalized_ang_vel,
state[16:20]
]).reshape(20,)
return norm_and_clipped
################################################################################
def _clipAndNormalizeStateWarning(self,
state,
clipped_pos_xy,
clipped_pos_z,
clipped_rp,
clipped_vel_xy,
clipped_vel_z,
):
"""Debugging printouts associated to `_clipAndNormalizeState`.
Print a warning if values in a state vector is out of the clipping range.
"""
if not(clipped_pos_xy == np.array(state[0:2])).all():
print("[WARNING] it", self.step_counter, "in FlockAviary._clipAndNormalizeState(), clipped xy position [{:.2f} {:.2f}]".format(state[0], state[1]))
if not(clipped_pos_z == np.array(state[2])).all():
print("[WARNING] it", self.step_counter, "in FlockAviary._clipAndNormalizeState(), clipped z position [{:.2f}]".format(state[2]))
if not(clipped_rp == np.array(state[7:9])).all():
print("[WARNING] it", self.step_counter, "in FlockAviary._clipAndNormalizeState(), clipped roll/pitch [{:.2f} {:.2f}]".format(state[7], state[8]))
if not(clipped_vel_xy == np.array(state[10:12])).all():
print("[WARNING] it", self.step_counter, "in FlockAviary._clipAndNormalizeState(), clipped xy velocity [{:.2f} {:.2f}]".format(state[10], state[11]))
if not(clipped_vel_z == np.array(state[12])).all():
print("[WARNING] it", self.step_counter, "in FlockAviary._clipAndNormalizeState(), clipped z velocity [{:.2f}]".format(state[12]))