-
Notifications
You must be signed in to change notification settings - Fork 8
/
walker3d.py
executable file
·354 lines (285 loc) · 14.4 KB
/
walker3d.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
__author__ = 'yuwenhao'
import numpy as np
from gym import utils
from gym.envs.dart import dart_env
import joblib
import os
import time
class DartWalker3dEnv(dart_env.DartEnv, utils.EzPickle):
def __init__(self):
self.control_bounds = np.array([[1.0] * 15, [-1.0] * 15])
self.action_scale = np.array([200.0, 200, 200, 250, 60, 80, 100, 60, 60, 250, 60, 80, 100, 60, 60])
obs_dim = 41
self.t = 0
self.target_vel = 1.0
self.init_tv = 0.0
self.final_tv = 1.0
self.tv_endtime = 0.5
self.alive_bonus = 4.0
self.smooth_tv_change = True
self.rand_target_vel = False
self.init_push = False
self.enforce_target_vel = True
self.running_avg_rew_only = True
self.avg_rew_weighting = []
self.vel_cache = []
self.reset_range = 0.05
self.target_ang = None
self.assist_timeout = 0.0
self.assist_prob = 1.0 # probability of providing assistance
self.assist_schedule = [[0.0, [2000, 2000]], [3.0, [1500, 1500]], [6.0, [1125.0, 1125.0]]]
self.hard_enforce = False
self.treadmill = False
self.treadmill_vel = -self.init_tv
self.treadmill_init_tv = -1.2
self.treadmill_final_tv = -1.2
self.treadmill_tv_endtime = 0.04
self.cur_step = 0
self.stepwise_rewards = []
self.conseq_limit_pen = 0 # number of steps lying on the wall
self.constrain_2d = True
self.init_balance_pd = 2000.0
self.init_vel_pd = 2000.0
self.current_pd = self.init_balance_pd
self.vel_enforce_kp = self.init_vel_pd
self.energy_weight = 0.3
self.vel_reward_weight = 3.0
self.foot_lift_weight = 5.0
self.local_spd_curriculum = True
self.anchor_kp = np.array([0, 0]) * 1.0
self.learns_turning = False
# state related
self.contact_info = np.array([0, 0])
self.include_additional_info = True
if self.include_additional_info:
obs_dim += len(self.contact_info)
if self.rand_target_vel or self.smooth_tv_change:
obs_dim += 1
self.curriculum_id = 0
self.spd_kp_candidates = None
if self.treadmill:
dart_env.DartEnv.__init__(self, 'walker3d_treadmill.skel', 15, obs_dim, self.control_bounds,
disableViewer=True)
else:
dart_env.DartEnv.__init__(self, 'walker3d_waist.skel', 15, obs_dim, self.control_bounds,
disableViewer=True, dt=0.002)
# self.dart_world.set_collision_detector(3) # uncomment if using ODE collision detector
self.robot_skeleton.set_self_collision_check(True)
for i in range(1, len(self.dart_world.skeletons[0].bodynodes)):
self.dart_world.skeletons[0].bodynodes[i].set_friction_coeff(0)
for i in range(0, len(self.dart_world.skeletons[0].bodynodes)):
self.dart_world.skeletons[0].bodynodes[i].set_friction_coeff(5)
for i in range(0, len(self.dart_world.skeletons[1].bodynodes)):
self.dart_world.skeletons[1].bodynodes[i].set_friction_coeff(5)
self.sim_dt = self.dt / self.frame_skip
for bn in self.robot_skeleton.bodynodes:
if len(bn.shapenodes) > 0:
shapesize = bn.shapenodes[0].shape.size()
print('density of ', bn.name, ' is ', bn.mass() / np.prod(shapesize))
print('Total mass: ', self.robot_skeleton.mass())
utils.EzPickle.__init__(self)
def _bodynode_spd(self, bn, kp, dof, target_vel=None):
self.Kp = kp
self.Kd = kp * self.sim_dt
if target_vel is not None:
self.Kd = self.Kp
self.Kp *= 0
invM = 1.0 / (bn.mass() + self.Kd * self.sim_dt)
p = -self.Kp * (bn.C[dof] + bn.dC[dof] * self.sim_dt)
if target_vel is None:
target_vel = 0.0
d = -self.Kd * (bn.dC[dof] - target_vel)
qddot = invM * (-bn.C[dof] + p + d)
tau = p + d - self.Kd * (qddot) * self.sim_dt
return tau
def do_simulation(self, tau, n_frames):
provide_assist = np.random.random() < self.assist_prob
for _ in range(n_frames):
if self.constrain_2d and self.t < self.assist_timeout and provide_assist:
force = self._bodynode_spd(self.robot_skeleton.bodynode('h_pelvis'), self.current_pd, 2)
self.robot_skeleton.bodynode('h_pelvis').add_ext_force(np.array([0, 0, force]))
if self.enforce_target_vel and not self.hard_enforce and self.t < self.assist_timeout and provide_assist:
force = self._bodynode_spd(self.robot_skeleton.bodynode('h_pelvis'), self.vel_enforce_kp, 0,
self.target_vel)
self.robot_skeleton.bodynode('h_pelvis').add_ext_force(np.array([force, 0, 0]))
if self.target_ang is not None:
tau[4] = self.vel_enforce_kp / 100 * (self.target_ang - self.robot_skeleton.q[4])
self.robot_skeleton.set_forces(tau)
self.dart_world.step()
s = self.state_vector()
if not (np.isfinite(s).all() and (np.abs(s[2:]) < 100).all()):
break
def advance(self, a):
clamped_control = np.array(a)
for i in range(len(clamped_control)):
if clamped_control[i] > self.control_bounds[0][i]:
clamped_control[i] = self.control_bounds[0][i]
if clamped_control[i] < self.control_bounds[1][i]:
clamped_control[i] = self.control_bounds[1][i]
tau = np.zeros(self.robot_skeleton.ndofs)
tau[6:] = clamped_control * self.action_scale
if self.enforce_target_vel:
if self.hard_enforce and self.treadmill:
current_dq_tread = self.dart_world.skeletons[0].dq
current_dq_tread[0] = self.treadmill_vel # * np.min([self.t/4.0, 1.0])
self.dart_world.skeletons[0].dq = current_dq_tread
elif self.hard_enforce:
current_dq = self.robot_skeleton.dq
current_dq[0] = self.target_vel
self.robot_skeleton.dq = current_dq
self.do_simulation(tau, self.frame_skip)
def _step(self, a):
if self.smooth_tv_change:
self.target_vel = (np.min([self.t, self.tv_endtime]) / self.tv_endtime) * (
self.final_tv - self.init_tv) + self.init_tv
self.treadmill_vel = (np.min([self.t, self.treadmill_tv_endtime]) / self.treadmill_tv_endtime) * (
self.treadmill_final_tv - self.treadmill_init_tv) + self.treadmill_init_tv
self.current_pd = self.init_balance_pd
self.vel_enforce_kp = self.init_vel_pd
if len(self.assist_schedule) > 0:
for sch in self.assist_schedule:
if self.t > sch[0]:
self.current_pd = sch[1][0]
self.vel_enforce_kp = sch[1][1]
pre_state = [self.state_vector()]
posbefore = self.robot_skeleton.bodynodes[0].com()[0]
self.advance(np.copy(a))
posafter = self.robot_skeleton.bodynodes[1].com()[0]
height = self.robot_skeleton.bodynodes[1].com()[1]
side_deviation = self.robot_skeleton.bodynodes[1].com()[2]
angle = self.robot_skeleton.q[3]
upward = np.array([0, 1, 0])
upward_world = self.robot_skeleton.bodynodes[1].to_world(np.array([0, 1, 0])) - self.robot_skeleton.bodynodes[
1].to_world(np.array([0, 0, 0]))
upward_world /= np.linalg.norm(upward_world)
ang_cos_uwd = np.dot(upward, upward_world)
ang_cos_uwd = np.arccos(ang_cos_uwd)
forward = np.array([1, 0, 0])
forward_world = self.robot_skeleton.bodynodes[1].to_world(np.array([1, 0, 0])) - self.robot_skeleton.bodynodes[
1].to_world(np.array([0, 0, 0]))
forward_world /= np.linalg.norm(forward_world)
ang_cos_fwd = np.dot(forward, forward_world)
ang_cos_fwd = np.arccos(ang_cos_fwd)
contacts = self.dart_world.collision_result.contacts
total_force_mag = 0
self_colliding = False
self.contact_info = np.array([0, 0])
l_foot_force = np.array([0.0, 0, 0])
r_foot_force = np.array([0.0, 0, 0])
for contact in contacts:
total_force_mag += np.square(contact.force).sum()
if contact.skel_id1 == contact.skel_id2:
self_colliding = True
if contact.skel_id1 + contact.skel_id2 == 1:
if contact.bodynode1 == self.robot_skeleton.bodynode('h_foot_left') or contact.bodynode2 == \
self.robot_skeleton.bodynode('h_foot_left'):
self.contact_info[0] = 1
l_foot_force += contact.force
if contact.bodynode1 == self.robot_skeleton.bodynode('h_foot') or contact.bodynode2 == \
self.robot_skeleton.bodynode('h_foot'):
self.contact_info[1] = 1
r_foot_force += contact.force
vel = (posafter - posbefore) / self.dt
self.vel_cache.append(vel)
self.target_vel_cache.append(self.target_vel)
if len(self.vel_cache) > int(2.0 / self.dt) and (self.running_avg_rew_only):
self.vel_cache.pop(0)
self.target_vel_cache.pop(0)
vel_rew = 0
if not self.treadmill:
if self.running_avg_rew_only:
vel_rew = -self.vel_reward_weight * np.abs(np.mean(self.target_vel_cache) - np.mean(self.vel_cache))
if self.t < self.tv_endtime:
vel_rew *= 0.5
else:
vel_rew = -self.vel_reward_weight * np.abs(self.target_vel - vel).sum()
else:
if self.running_avg_rew_only:
append_vel = np.ones(int(1.0 / self.dt) - len(self.vel_cache)) * (self.target_vel + self.treadmill_vel)
vel_rew = -3.0 * (
np.abs(self.target_vel + self.treadmill_vel - np.mean(np.append(self.vel_cache, append_vel))))
else:
vel_rew = -3.0 * (np.abs(self.target_vel + self.treadmill_vel - vel))
if self.t < self.tv_endtime:
vel_rew *= 1.0
if self.target_ang is not None:
print(self.robot_skeleton.q[4])
ang_vel_rew = - 3.0 * np.abs(np.abs(self.target_ang) - np.abs(self.robot_skeleton.q[4]))
else:
ang_vel_rew = 0
action_pen = self.energy_weight * np.abs(a).sum()
deviation_pen = 3 * abs(side_deviation)
rot_pen = 1.0 * np.abs(self.robot_skeleton.q[3]) + 0.0 * np.abs(self.robot_skeleton.q[4]) + \
1.0 * np.abs(self.robot_skeleton.q[5])
jump_rew = 10.0 * np.max([(height - 1.3), 0])
foot_rew = self.foot_lift_weight * (np.max(
[self.robot_skeleton.bodynode('h_thigh').C[1], self.robot_skeleton.bodynode('h_thigh_left').C[1]]) - 0.8)
reward = vel_rew + self.alive_bonus - action_pen - deviation_pen - rot_pen + foot_rew # + jump_rew + ang_vel_rew# - contact_pen
pos_rew = self.alive_bonus - deviation_pen
neg_pen = vel_rew - action_pen
self.t += self.dt
self.cur_step += 1
s = self.state_vector()
done = not (np.isfinite(s).all() and (np.abs(s[2:]) < 100).all() and
(height > 1.0) and (height < 3.0) and (abs(ang_cos_uwd) < 1.2) and (
abs(ang_cos_fwd) < 1.2)
and np.abs(angle) < 1.1 and np.abs(self.robot_skeleton.q[5]) < 1.2 and np.abs(
side_deviation) < 0.9)
self.stepwise_rewards.append(reward)
broke_sim = False
if not (np.isfinite(s).all() and (np.abs(s[2:]) < 100).all()):
broke_sim = True
if broke_sim:
reward = 0
ob = self._get_obs()
return ob, reward, done, {'pos_rew': pos_rew, 'neg_pen': neg_pen, 'broke_sim': broke_sim,
'pre_state': pre_state,
'vel_rew': vel_rew, 'action_pen': action_pen / self.energy_weight,
'deviation_pen': deviation_pen,
'curriculum_id': self.curriculum_id, 'curriculum_candidates': self.spd_kp_candidates,
'done_return': done, 'dyn_model_id': 0, 'state_index': 0,
'contact_forces': [l_foot_force, r_foot_force],
'contact_force': l_foot_force + r_foot_force, 'avg_vel': np.mean(self.vel_cache)}
def _get_obs(self):
state = np.concatenate([
self.robot_skeleton.q[1:],
self.robot_skeleton.dq,
])
if self.include_additional_info:
state = np.concatenate([state, self.contact_info])
if self.rand_target_vel or self.smooth_tv_change:
state = np.concatenate([state, [self.target_vel]])
return state
def reset_model(self):
self.dart_world.reset()
qpos = self.robot_skeleton.q + self.np_random.uniform(low=-self.reset_range, high=self.reset_range, size=self.robot_skeleton.ndofs)
qvel = self.robot_skeleton.dq + self.np_random.uniform(low=-self.reset_range, high=self.reset_range, size=self.robot_skeleton.ndofs)
if self.rand_target_vel:
self.target_vel = np.random.uniform(0.8, 2.5)
if self.local_spd_curriculum:
self.spd_kp_candidates = [self.anchor_kp]
self.curriculum_id = np.random.randint(len(self.spd_kp_candidates))
chosen_curriculum = self.spd_kp_candidates[self.curriculum_id]
self.init_balance_pd = chosen_curriculum[0]
self.init_vel_pd = chosen_curriculum[1]
qpos[3] -= 0.05
if self.init_push:
qvel[0] = self.target_vel
self.set_state(qpos, qvel)
self.t = 0
self.cur_step = 0
self.stepwise_rewards = []
self.init_pos = self.robot_skeleton.q[0]
self.vel_cache = []
self.target_vel_cache = []
self.avg_rew_weighting = []
self.conseq_limit_pen = 0
self.current_pd = self.init_balance_pd
self.vel_enforce_kp = self.init_vel_pd
self.contact_info = np.array([0, 0])
if self.target_ang is not None and np.random.random() < 0.5:
self.target_ang *= -1
return self._get_obs()
def viewer_setup(self):
if not self.disableViewer:
self._get_viewer().scene.tb.trans[2] = -5.5