-
Notifications
You must be signed in to change notification settings - Fork 443
/
Copy pathfactory_base.py
536 lines (458 loc) · 31.3 KB
/
factory_base.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
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
# Copyright (c) 2021-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: base class.
Inherits Gym's VecTask class and abstract base class. Inherited by environment classes. Not directly executed.
Configuration defined in FactoryBase.yaml. Asset info defined in factory_asset_info_franka_table.yaml.
"""
import hydra
import math
import numpy as np
import os
import sys
import torch
from gym import logger
from isaacgym import gymapi, gymtorch
from isaacgymenvs.utils import torch_jit_utils as torch_utils
from isaacgymenvs.tasks.base.vec_task import VecTask
import isaacgymenvs.tasks.factory.factory_control as fc
from isaacgymenvs.tasks.factory.factory_schema_class_base import FactoryABCBase
from isaacgymenvs.tasks.factory.factory_schema_config_base import FactorySchemaConfigBase
class FactoryBase(VecTask, FactoryABCBase):
def __init__(self, cfg, rl_device, sim_device, graphics_device_id, headless, virtual_screen_capture, force_render):
"""Initialize instance variables. Initialize VecTask superclass."""
self.cfg = cfg
self.cfg['headless'] = headless
self._get_base_yaml_params()
if self.cfg_base.mode.export_scene:
sim_device = 'cpu'
super().__init__(cfg, rl_device, sim_device, graphics_device_id, headless, virtual_screen_capture, force_render) # create_sim() is called here
def _get_base_yaml_params(self):
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name='factory_schema_config_base', node=FactorySchemaConfigBase)
config_path = 'task/FactoryBase.yaml' # relative to Gym's Hydra search path (cfg dir)
self.cfg_base = hydra.compose(config_name=config_path)
self.cfg_base = self.cfg_base['task'] # strip superfluous nesting
asset_info_path = '../../assets/factory/yaml/factory_asset_info_franka_table.yaml' # relative to Gym's Hydra search path (cfg dir)
self.asset_info_franka_table = hydra.compose(config_name=asset_info_path)
self.asset_info_franka_table = self.asset_info_franka_table['']['']['']['']['']['']['assets']['factory']['yaml'] # strip superfluous nesting
def create_sim(self):
"""Set sim and PhysX params. Create sim object, ground plane, and envs."""
if self.cfg_base.mode.export_scene:
self.sim_params.use_gpu_pipeline = False
self.sim = super().create_sim(compute_device=self.device_id,
graphics_device=self.graphics_device_id,
physics_engine=self.physics_engine,
sim_params=self.sim_params)
self._create_ground_plane()
self.create_envs() # defined in subclass
def _create_ground_plane(self):
"""Set ground plane params. Add plane."""
plane_params = gymapi.PlaneParams()
plane_params.normal = gymapi.Vec3(0.0, 0.0, 1.0)
plane_params.distance = 0.0 # default = 0.0
plane_params.static_friction = 1.0 # default = 1.0
plane_params.dynamic_friction = 1.0 # default = 1.0
plane_params.restitution = 0.0 # default = 0.0
self.gym.add_ground(self.sim, plane_params)
def import_franka_assets(self):
"""Set Franka and table asset options. Import assets."""
urdf_root = os.path.join(os.path.dirname(__file__), '..', '..', '..', 'assets', 'factory', 'urdf')
franka_file = 'factory_franka.urdf'
franka_options = gymapi.AssetOptions()
franka_options.flip_visual_attachments = True
franka_options.fix_base_link = True
franka_options.collapse_fixed_joints = False
franka_options.thickness = 0.0 # default = 0.02
franka_options.density = 1000.0 # default = 1000.0
franka_options.armature = 0.01 # default = 0.0
franka_options.use_physx_armature = True
if self.cfg_base.sim.add_damping:
franka_options.linear_damping = 1.0 # default = 0.0; increased to improve stability
franka_options.max_linear_velocity = 1.0 # default = 1000.0; reduced to prevent CUDA errors
franka_options.angular_damping = 5.0 # default = 0.5; increased to improve stability
franka_options.max_angular_velocity = 2 * math.pi # default = 64.0; reduced to prevent CUDA errors
else:
franka_options.linear_damping = 0.0 # default = 0.0
franka_options.max_linear_velocity = 1000.0 # default = 1000.0
franka_options.angular_damping = 0.5 # default = 0.5
franka_options.max_angular_velocity = 64.0 # default = 64.0
franka_options.disable_gravity = True
franka_options.enable_gyroscopic_forces = True
franka_options.default_dof_drive_mode = gymapi.DOF_MODE_NONE
franka_options.use_mesh_materials = True
if self.cfg_base.mode.export_scene:
franka_options.mesh_normal_mode = gymapi.COMPUTE_PER_FACE
table_options = gymapi.AssetOptions()
table_options.flip_visual_attachments = False # default = False
table_options.fix_base_link = True
table_options.thickness = 0.0 # default = 0.02
table_options.density = 1000.0 # default = 1000.0
table_options.armature = 0.0 # default = 0.0
table_options.use_physx_armature = True
table_options.linear_damping = 0.0 # default = 0.0
table_options.max_linear_velocity = 1000.0 # default = 1000.0
table_options.angular_damping = 0.0 # default = 0.5
table_options.max_angular_velocity = 64.0 # default = 64.0
table_options.disable_gravity = False
table_options.enable_gyroscopic_forces = True
table_options.default_dof_drive_mode = gymapi.DOF_MODE_NONE
table_options.use_mesh_materials = False
if self.cfg_base.mode.export_scene:
table_options.mesh_normal_mode = gymapi.COMPUTE_PER_FACE
franka_asset = self.gym.load_asset(self.sim, urdf_root, franka_file, franka_options)
table_asset = self.gym.create_box(self.sim, self.asset_info_franka_table.table_depth,
self.asset_info_franka_table.table_width, self.cfg_base.env.table_height,
table_options)
return franka_asset, table_asset
def acquire_base_tensors(self):
"""Acquire and wrap tensors. Create views."""
_root_state = self.gym.acquire_actor_root_state_tensor(self.sim) # shape = (num_envs * num_actors, 13)
_body_state = self.gym.acquire_rigid_body_state_tensor(self.sim) # shape = (num_envs * num_bodies, 13)
_dof_state = self.gym.acquire_dof_state_tensor(self.sim) # shape = (num_envs * num_dofs, 2)
_dof_force = self.gym.acquire_dof_force_tensor(self.sim) # shape = (num_envs * num_dofs, 1)
_contact_force = self.gym.acquire_net_contact_force_tensor(self.sim) # shape = (num_envs * num_bodies, 3)
_jacobian = self.gym.acquire_jacobian_tensor(self.sim, 'franka') # shape = (num envs, num_bodies, 6, num_dofs)
_mass_matrix = self.gym.acquire_mass_matrix_tensor(self.sim, 'franka') # shape = (num_envs, num_dofs, num_dofs)
self.root_state = gymtorch.wrap_tensor(_root_state)
self.body_state = gymtorch.wrap_tensor(_body_state)
self.dof_state = gymtorch.wrap_tensor(_dof_state)
self.dof_force = gymtorch.wrap_tensor(_dof_force)
self.contact_force = gymtorch.wrap_tensor(_contact_force)
self.jacobian = gymtorch.wrap_tensor(_jacobian)
self.mass_matrix = gymtorch.wrap_tensor(_mass_matrix)
self.root_pos = self.root_state.view(self.num_envs, self.num_actors, 13)[..., 0:3]
self.root_quat = self.root_state.view(self.num_envs, self.num_actors, 13)[..., 3:7]
self.root_linvel = self.root_state.view(self.num_envs, self.num_actors, 13)[..., 7:10]
self.root_angvel = self.root_state.view(self.num_envs, self.num_actors, 13)[..., 10:13]
self.body_pos = self.body_state.view(self.num_envs, self.num_bodies, 13)[..., 0:3]
self.body_quat = self.body_state.view(self.num_envs, self.num_bodies, 13)[..., 3:7]
self.body_linvel = self.body_state.view(self.num_envs, self.num_bodies, 13)[..., 7:10]
self.body_angvel = self.body_state.view(self.num_envs, self.num_bodies, 13)[..., 10:13]
self.dof_pos = self.dof_state.view(self.num_envs, self.num_dofs, 2)[..., 0]
self.dof_vel = self.dof_state.view(self.num_envs, self.num_dofs, 2)[..., 1]
self.dof_force_view = self.dof_force.view(self.num_envs, self.num_dofs, 1)[..., 0]
self.contact_force = self.contact_force.view(self.num_envs, self.num_bodies, 3)[..., 0:3]
self.arm_dof_pos = self.dof_pos[:, 0:7]
self.arm_mass_matrix = self.mass_matrix[:, 0:7, 0:7] # for Franka arm (not gripper)
self.hand_pos = self.body_pos[:, self.hand_body_id_env, 0:3]
self.hand_quat = self.body_quat[:, self.hand_body_id_env, 0:4]
self.hand_linvel = self.body_linvel[:, self.hand_body_id_env, 0:3]
self.hand_angvel = self.body_angvel[:, self.hand_body_id_env, 0:3]
self.hand_jacobian = self.jacobian[:, self.hand_body_id_env - 1, 0:6, 0:7] # minus 1 because base is fixed
self.left_finger_pos = self.body_pos[:, self.left_finger_body_id_env, 0:3]
self.left_finger_quat = self.body_quat[:, self.left_finger_body_id_env, 0:4]
self.left_finger_linvel = self.body_linvel[:, self.left_finger_body_id_env, 0:3]
self.left_finger_angvel = self.body_angvel[:, self.left_finger_body_id_env, 0:3]
self.left_finger_jacobian = self.jacobian[:, self.left_finger_body_id_env - 1, 0:6, 0:7] # minus 1 because base is fixed
self.right_finger_pos = self.body_pos[:, self.right_finger_body_id_env, 0:3]
self.right_finger_quat = self.body_quat[:, self.right_finger_body_id_env, 0:4]
self.right_finger_linvel = self.body_linvel[:, self.right_finger_body_id_env, 0:3]
self.right_finger_angvel = self.body_angvel[:, self.right_finger_body_id_env, 0:3]
self.right_finger_jacobian = self.jacobian[:, self.right_finger_body_id_env - 1, 0:6, 0:7] # minus 1 because base is fixed
self.left_finger_force = self.contact_force[:, self.left_finger_body_id_env, 0:3]
self.right_finger_force = self.contact_force[:, self.right_finger_body_id_env, 0:3]
self.gripper_dof_pos = self.dof_pos[:, 7:9]
self.fingertip_centered_pos = self.body_pos[:, self.fingertip_centered_body_id_env, 0:3]
self.fingertip_centered_quat = self.body_quat[:, self.fingertip_centered_body_id_env, 0:4]
self.fingertip_centered_linvel = self.body_linvel[:, self.fingertip_centered_body_id_env, 0:3]
self.fingertip_centered_angvel = self.body_angvel[:, self.fingertip_centered_body_id_env, 0:3]
self.fingertip_centered_jacobian = self.jacobian[:, self.fingertip_centered_body_id_env - 1, 0:6, 0:7] # minus 1 because base is fixed
self.fingertip_midpoint_pos = self.fingertip_centered_pos.detach().clone() # initial value
self.fingertip_midpoint_quat = self.fingertip_centered_quat # always equal
self.fingertip_midpoint_linvel = self.fingertip_centered_linvel.detach().clone() # initial value
# From sum of angular velocities (https://physics.stackexchange.com/questions/547698/understanding-addition-of-angular-velocity),
# angular velocity of midpoint w.r.t. world is equal to sum of
# angular velocity of midpoint w.r.t. hand and angular velocity of hand w.r.t. world.
# Midpoint is in sliding contact (i.e., linear relative motion) with hand; angular velocity of midpoint w.r.t. hand is zero.
# Thus, angular velocity of midpoint w.r.t. world is equal to angular velocity of hand w.r.t. world.
self.fingertip_midpoint_angvel = self.fingertip_centered_angvel # always equal
self.fingertip_midpoint_jacobian = (self.left_finger_jacobian + self.right_finger_jacobian) * 0.5 # approximation
self.dof_torque = torch.zeros((self.num_envs, self.num_dofs), device=self.device)
self.fingertip_contact_wrench = torch.zeros((self.num_envs, 6), device=self.device)
self.ctrl_target_fingertip_midpoint_pos = torch.zeros((self.num_envs, 3), device=self.device)
self.ctrl_target_fingertip_midpoint_quat = torch.zeros((self.num_envs, 4), device=self.device)
self.ctrl_target_dof_pos = torch.zeros((self.num_envs, self.num_dofs), device=self.device)
self.ctrl_target_gripper_dof_pos = torch.zeros((self.num_envs, 2), device=self.device)
self.ctrl_target_fingertip_contact_wrench = torch.zeros((self.num_envs, 6), device=self.device)
self.prev_actions = torch.zeros((self.num_envs, self.num_actions), device=self.device)
def refresh_base_tensors(self):
"""Refresh tensors."""
# NOTE: Tensor refresh functions should be called once per step, before setters.
self.gym.refresh_dof_state_tensor(self.sim)
self.gym.refresh_actor_root_state_tensor(self.sim)
self.gym.refresh_rigid_body_state_tensor(self.sim)
self.gym.refresh_dof_force_tensor(self.sim)
self.gym.refresh_net_contact_force_tensor(self.sim)
self.gym.refresh_jacobian_tensors(self.sim)
self.gym.refresh_mass_matrix_tensors(self.sim)
self.finger_midpoint_pos = (self.left_finger_pos + self.right_finger_pos) * 0.5
self.fingertip_midpoint_pos = fc.translate_along_local_z(pos=self.finger_midpoint_pos,
quat=self.hand_quat,
offset=self.asset_info_franka_table.franka_finger_length,
device=self.device)
# TODO: Add relative velocity term (see https://dynamicsmotioncontrol487379916.files.wordpress.com/2020/11/21-me258pointmovingrigidbody.pdf)
self.fingertip_midpoint_linvel = self.fingertip_centered_linvel + torch.cross(self.fingertip_centered_angvel,
(self.fingertip_midpoint_pos - self.fingertip_centered_pos),
dim=1)
self.fingertip_midpoint_jacobian = (self.left_finger_jacobian + self.right_finger_jacobian) * 0.5 # approximation
def parse_controller_spec(self):
"""Parse controller specification into lower-level controller configuration."""
cfg_ctrl_keys = {'num_envs',
'jacobian_type',
'gripper_prop_gains',
'gripper_deriv_gains',
'motor_ctrl_mode',
'gain_space',
'ik_method',
'joint_prop_gains',
'joint_deriv_gains',
'do_motion_ctrl',
'task_prop_gains',
'task_deriv_gains',
'do_inertial_comp',
'motion_ctrl_axes',
'do_force_ctrl',
'force_ctrl_method',
'wrench_prop_gains',
'force_ctrl_axes'}
self.cfg_ctrl = {cfg_ctrl_key: None for cfg_ctrl_key in cfg_ctrl_keys}
self.cfg_ctrl['num_envs'] = self.num_envs
self.cfg_ctrl['jacobian_type'] = self.cfg_task.ctrl.all.jacobian_type
self.cfg_ctrl['gripper_prop_gains'] = torch.tensor(self.cfg_task.ctrl.all.gripper_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['gripper_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.all.gripper_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
ctrl_type = self.cfg_task.ctrl.ctrl_type
if ctrl_type == 'gym_default':
self.cfg_ctrl['motor_ctrl_mode'] = 'gym'
self.cfg_ctrl['gain_space'] = 'joint'
self.cfg_ctrl['ik_method'] = self.cfg_task.ctrl.gym_default.ik_method
self.cfg_ctrl['joint_prop_gains'] = torch.tensor(self.cfg_task.ctrl.gym_default.joint_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['joint_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.gym_default.joint_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['gripper_prop_gains'] = torch.tensor(self.cfg_task.ctrl.gym_default.gripper_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['gripper_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.gym_default.gripper_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
elif ctrl_type == 'joint_space_ik':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'joint'
self.cfg_ctrl['ik_method'] = self.cfg_task.ctrl.joint_space_ik.ik_method
self.cfg_ctrl['joint_prop_gains'] = torch.tensor(self.cfg_task.ctrl.joint_space_ik.joint_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['joint_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.joint_space_ik.joint_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_inertial_comp'] = False
elif ctrl_type == 'joint_space_id':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'joint'
self.cfg_ctrl['ik_method'] = self.cfg_task.ctrl.joint_space_id.ik_method
self.cfg_ctrl['joint_prop_gains'] = torch.tensor(self.cfg_task.ctrl.joint_space_id.joint_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['joint_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.joint_space_id.joint_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_inertial_comp'] = True
elif ctrl_type == 'task_space_impedance':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'task'
self.cfg_ctrl['do_motion_ctrl'] = True
self.cfg_ctrl['task_prop_gains'] = torch.tensor(self.cfg_task.ctrl.task_space_impedance.task_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['task_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.task_space_impedance.task_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_inertial_comp'] = False
self.cfg_ctrl['motion_ctrl_axes'] = torch.tensor(self.cfg_task.ctrl.task_space_impedance.motion_ctrl_axes,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_force_ctrl'] = False
elif ctrl_type == 'operational_space_motion':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'task'
self.cfg_ctrl['do_motion_ctrl'] = True
self.cfg_ctrl['task_prop_gains'] = torch.tensor(self.cfg_task.ctrl.operational_space_motion.task_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['task_deriv_gains'] = torch.tensor(
self.cfg_task.ctrl.operational_space_motion.task_deriv_gains, device=self.device).repeat(
(self.num_envs, 1))
self.cfg_ctrl['do_inertial_comp'] = True
self.cfg_ctrl['motion_ctrl_axes'] = torch.tensor(
self.cfg_task.ctrl.operational_space_motion.motion_ctrl_axes, device=self.device).repeat(
(self.num_envs, 1))
self.cfg_ctrl['do_force_ctrl'] = False
elif ctrl_type == 'open_loop_force':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'task'
self.cfg_ctrl['do_motion_ctrl'] = False
self.cfg_ctrl['do_force_ctrl'] = True
self.cfg_ctrl['force_ctrl_method'] = 'open'
self.cfg_ctrl['force_ctrl_axes'] = torch.tensor(self.cfg_task.ctrl.open_loop_force.force_ctrl_axes,
device=self.device).repeat((self.num_envs, 1))
elif ctrl_type == 'closed_loop_force':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'task'
self.cfg_ctrl['do_motion_ctrl'] = False
self.cfg_ctrl['do_force_ctrl'] = True
self.cfg_ctrl['force_ctrl_method'] = 'closed'
self.cfg_ctrl['wrench_prop_gains'] = torch.tensor(self.cfg_task.ctrl.closed_loop_force.wrench_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['force_ctrl_axes'] = torch.tensor(self.cfg_task.ctrl.closed_loop_force.force_ctrl_axes,
device=self.device).repeat((self.num_envs, 1))
elif ctrl_type == 'hybrid_force_motion':
self.cfg_ctrl['motor_ctrl_mode'] = 'manual'
self.cfg_ctrl['gain_space'] = 'task'
self.cfg_ctrl['do_motion_ctrl'] = True
self.cfg_ctrl['task_prop_gains'] = torch.tensor(self.cfg_task.ctrl.hybrid_force_motion.task_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['task_deriv_gains'] = torch.tensor(self.cfg_task.ctrl.hybrid_force_motion.task_deriv_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_inertial_comp'] = True
self.cfg_ctrl['motion_ctrl_axes'] = torch.tensor(self.cfg_task.ctrl.hybrid_force_motion.motion_ctrl_axes,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['do_force_ctrl'] = True
self.cfg_ctrl['force_ctrl_method'] = 'closed'
self.cfg_ctrl['wrench_prop_gains'] = torch.tensor(self.cfg_task.ctrl.hybrid_force_motion.wrench_prop_gains,
device=self.device).repeat((self.num_envs, 1))
self.cfg_ctrl['force_ctrl_axes'] = torch.tensor(self.cfg_task.ctrl.hybrid_force_motion.force_ctrl_axes,
device=self.device).repeat((self.num_envs, 1))
if self.cfg_ctrl['motor_ctrl_mode'] == 'gym':
prop_gains = torch.cat((self.cfg_ctrl['joint_prop_gains'],
self.cfg_ctrl['gripper_prop_gains']), dim=-1).to('cpu')
deriv_gains = torch.cat((self.cfg_ctrl['joint_deriv_gains'],
self.cfg_ctrl['gripper_deriv_gains']), dim=-1).to('cpu')
# No tensor API for getting/setting actor DOF props; thus, loop required
for env_ptr, franka_handle, prop_gain, deriv_gain in zip(self.env_ptrs, self.franka_handles, prop_gains,
deriv_gains):
franka_dof_props = self.gym.get_actor_dof_properties(env_ptr, franka_handle)
franka_dof_props['driveMode'][:] = gymapi.DOF_MODE_POS
franka_dof_props['stiffness'] = prop_gain
franka_dof_props['damping'] = deriv_gain
self.gym.set_actor_dof_properties(env_ptr, franka_handle, franka_dof_props)
elif self.cfg_ctrl['motor_ctrl_mode'] == 'manual':
# No tensor API for getting/setting actor DOF props; thus, loop required
for env_ptr, franka_handle in zip(self.env_ptrs, self.franka_handles):
franka_dof_props = self.gym.get_actor_dof_properties(env_ptr, franka_handle)
franka_dof_props['driveMode'][:] = gymapi.DOF_MODE_EFFORT
franka_dof_props['stiffness'][:] = 0.0 # zero passive stiffness
franka_dof_props['damping'][:] = 0.0 # zero passive damping
self.gym.set_actor_dof_properties(env_ptr, franka_handle, franka_dof_props)
def generate_ctrl_signals(self):
"""Get Jacobian. Set Franka DOF position targets or DOF torques."""
# Get desired Jacobian
if self.cfg_ctrl['jacobian_type'] == 'geometric':
self.fingertip_midpoint_jacobian_tf = self.fingertip_midpoint_jacobian
elif self.cfg_ctrl['jacobian_type'] == 'analytic':
self.fingertip_midpoint_jacobian_tf = fc.get_analytic_jacobian(
fingertip_quat=self.fingertip_quat,
fingertip_jacobian=self.fingertip_midpoint_jacobian,
num_envs=self.num_envs,
device=self.device)
# Set PD joint pos target or joint torque
if self.cfg_ctrl['motor_ctrl_mode'] == 'gym':
self._set_dof_pos_target()
elif self.cfg_ctrl['motor_ctrl_mode'] == 'manual':
self._set_dof_torque()
def _set_dof_pos_target(self):
"""Set Franka DOF position target to move fingertips towards target pose."""
self.ctrl_target_dof_pos = fc.compute_dof_pos_target(
cfg_ctrl=self.cfg_ctrl,
arm_dof_pos=self.arm_dof_pos,
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
jacobian=self.fingertip_midpoint_jacobian_tf,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
ctrl_target_gripper_dof_pos=self.ctrl_target_gripper_dof_pos,
device=self.device)
self.gym.set_dof_position_target_tensor_indexed(self.sim,
gymtorch.unwrap_tensor(self.ctrl_target_dof_pos),
gymtorch.unwrap_tensor(self.franka_actor_ids_sim),
len(self.franka_actor_ids_sim))
def _set_dof_torque(self):
"""Set Franka DOF torque to move fingertips towards target pose."""
self.dof_torque = fc.compute_dof_torque(
cfg_ctrl=self.cfg_ctrl,
dof_pos=self.dof_pos,
dof_vel=self.dof_vel,
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
fingertip_midpoint_linvel=self.fingertip_midpoint_linvel,
fingertip_midpoint_angvel=self.fingertip_midpoint_angvel,
left_finger_force=self.left_finger_force,
right_finger_force=self.right_finger_force,
jacobian=self.fingertip_midpoint_jacobian_tf,
arm_mass_matrix=self.arm_mass_matrix,
ctrl_target_gripper_dof_pos=self.ctrl_target_gripper_dof_pos,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
ctrl_target_fingertip_contact_wrench=self.ctrl_target_fingertip_contact_wrench,
device=self.device)
self.gym.set_dof_actuation_force_tensor_indexed(self.sim,
gymtorch.unwrap_tensor(self.dof_torque),
gymtorch.unwrap_tensor(self.franka_actor_ids_sim),
len(self.franka_actor_ids_sim))
def print_sdf_warning(self):
"""Generate SDF warning message."""
logger.warn('Please be patient: SDFs may be generating, which may take a few minutes. Terminating prematurely may result in a corrupted SDF cache.')
def enable_gravity(self, gravity_mag):
"""Enable gravity."""
sim_params = self.gym.get_sim_params(self.sim)
sim_params.gravity.z = -gravity_mag
self.gym.set_sim_params(self.sim, sim_params)
def disable_gravity(self):
"""Disable gravity."""
sim_params = self.gym.get_sim_params(self.sim)
sim_params.gravity.z = 0.0
self.gym.set_sim_params(self.sim, sim_params)
def export_scene(self, label):
"""Export scene to USD."""
usd_export_options = gymapi.UsdExportOptions()
usd_export_options.export_physics = False
usd_exporter = self.gym.create_usd_exporter(usd_export_options)
self.gym.export_usd_sim(usd_exporter, self.sim, label)
sys.exit()
def extract_poses(self):
"""Extract poses of all bodies."""
if not hasattr(self, 'export_pos'):
self.export_pos = []
self.export_rot = []
self.frame_count = 0
pos = self.body_pos
rot = self.body_quat
self.export_pos.append(pos.cpu().numpy().copy())
self.export_rot.append(rot.cpu().numpy().copy())
self.frame_count += 1
if len(self.export_pos) == self.max_episode_length:
output_dir = self.__class__.__name__
save_dir = os.path.join('usd', output_dir)
os.makedirs(output_dir, exist_ok=True)
print(f'Exporting poses to {output_dir}...')
np.save(os.path.join(save_dir, 'body_position.npy'), np.array(self.export_pos))
np.save(os.path.join(save_dir, 'body_rotation.npy'), np.array(self.export_rot))
print('Export completed.')
sys.exit()