-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathrun_hpc.py
198 lines (177 loc) · 8.14 KB
/
run_hpc.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
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Tue Jan 24 13:11:51 2023
@author: orochi
"""
from multiprocessing import connection
from operator import truediv
import pybullet as p
import pybullet_data
import pathlib
from demos.rl_demo import rl_env
from demos.rl_demo import manipulation_phase_rl
# import rl_env
from demos.rl_demo.rl_state import StateRL, GoalHolder
from demos.rl_demo import rl_action
from demos.rl_demo import rl_reward
import pandas as pd
from mojograsp.simcore.sim_manager_HER import SimManagerRLHER
from mojograsp.simcore.state import StateDefault
from mojograsp.simcore.reward import RewardDefault
from mojograsp.simcore.record_data import RecordDataJSON, RecordDataPKL, RecordDataRLPKL
from mojograsp.simobjects.two_finger_gripper import TwoFingerGripper
from mojograsp.simobjects.object_base import ObjectBase
from mojograsp.simobjects.object_with_velocity import ObjectWithVelocity
from mojograsp.simobjects.object_for_dataframe import ObjectVelocityDF
from mojograsp.simcore.replay_buffer import ReplayBufferDefault, ReplayBufferDF
from mojograsp.simcore.episode import EpisodeDefault
import numpy as np
from mojograsp.simcore.priority_replay_buffer import ReplayBufferPriority
from mojograsp.simcore.data_combination import data_processor
import pickle as pkl
import json
import time
import sys
import os
def run_pybullet(filepath, window=None, runtype='run'):
# resource paths
with open(filepath, 'r') as argfile:
args = json.load(argfile)
if (args['task'] == 'asterisk'):
x = [0.03, 0, -0.03, -0.04, -0.03, 0, 0.03, 0.04]
y = [-0.03, -0.04, -0.03, 0, 0.03, 0.04, 0.03, 0]
elif args['task'] == 'random' and runtype != 'eval':
df = pd.read_csv(args['points_path'], index_col=False)
x = df["x"]
y = df["y"]
elif runtype=='eval':
# df = pd.read_csv('/home/orochi/mojo/mojo-grasp/demos/rl_demo/resources/test_points.csv', index_col=False)
# print('EVALUATING BOOOIIII')
# x = df["x"]
# y = df["y"]
x = [0.03, 0, -0.03, -0.04, -0.03, 0, 0.03, 0.04]
y = [-0.03, -0.04, -0.03, 0, 0.03, 0.04, 0.03, 0]
pose_list = [[i,j] for i,j in zip(x,y)]
print(args)
try:
if (args['viz']) | (runtype=='eval'):
physics_client = p.connect(p.GUI)
else:
physics_client = p.connect(p.DIRECT)
except KeyError:
physics_client = p.connect(p.GUI)
p.setAdditionalSearchPath(pybullet_data.getDataPath())
p.setGravity(0, 0, -10)
p.resetDebugVisualizerCamera(cameraDistance=.02, cameraYaw=0, cameraPitch=-89.9999,
cameraTargetPosition=[0, 0.1, 0.5])
# load objects into pybullet
plane_id = p.loadURDF("plane.urdf", flags=p.URDF_ENABLE_CACHED_GRAPHICS_SHAPES)
hand_id = p.loadURDF(args['hand_path'], useFixedBase=True,
basePosition=[0.0, 0.0, 0.05], flags=p.URDF_ENABLE_CACHED_GRAPHICS_SHAPES)
obj_id = p.loadURDF(args['object_path'], basePosition=[0.0, 0.10, .05], flags=p.URDF_ENABLE_CACHED_GRAPHICS_SHAPES)
# Create TwoFingerGripper Object and set the initial joint positions
hand = TwoFingerGripper(hand_id, path=args['hand_path'])
# p.resetJointState(hand_id, 0, -0.4)
# p.resetJointState(hand_id, 1, 1.2)
# p.resetJointState(hand_id, 3, 0.4)
# p.resetJointState(hand_id, 4, -1.2)
# p.resetJointState(hand_id, 0, 0)
# p.resetJointState(hand_id, 1, 0)
# p.resetJointState(hand_id, 3, 0)
# p.resetJointState(hand_id, 4, 0)
# change visual of gripper
p.changeVisualShape(hand_id, 0, rgbaColor=[0.3, 0.3, 0.3, 1])
p.changeVisualShape(hand_id, 1, rgbaColor=[1, 0.5, 0, 1])
p.changeVisualShape(hand_id, 3, rgbaColor=[0.3, 0.3, 0.3, 1])
p.changeVisualShape(hand_id, 4, rgbaColor=[1, 0.5, 0, 1])
p.changeVisualShape(hand_id, -1, rgbaColor=[0.3, 0.3, 0.3, 1])
p.configureDebugVisualizer(p.COV_ENABLE_RENDERING,0)
# p.setTimeStep(1/2400)
obj = ObjectWithVelocity(obj_id, path=args['object_path'])
# p.addUserDebugPoints([[0.2,0.1,0.0],[1,0,0]],[[1,0.0,0],[0.5,0.5,0.5]], 1)
p.configureDebugVisualizer(p.COV_ENABLE_RENDERING, 1)
goal_poses = GoalHolder(pose_list)
# time.sleep(10)
# state, action and reward
state = StateRL(objects=[hand, obj, goal_poses], prev_len=args['pv'])
action = rl_action.ExpertAction()
reward = rl_reward.ExpertReward()
#argument preprocessing
arg_dict = args.copy()
if args['action'] == 'Joint Velocity':
arg_dict['ik_flag'] = False
else:
arg_dict['ik_flag'] = True
# replay buffer
replay_buffer = ReplayBufferPriority(buffer_size=4080000)
# environment and recording
env = rl_env.ExpertEnv(hand=hand, obj=obj)
# env = rl_env.ExpertEnv(hand=hand, obj=cylinder)
# Create phase
manipulation = manipulation_phase_rl.ManipulationRL(
hand, obj, x, y, state, action, reward, replay_buffer=replay_buffer, args=arg_dict)
# data recording
record_data = RecordDataRLPKL(
data_path=args['save_path'], state=state, action=action, reward=reward, save_all=False, controller=manipulation.controller)
# sim manager
manager = SimManagerRLHER(num_episodes=len(pose_list), env=env, episode=EpisodeDefault(), record_data=record_data, replay_buffer=replay_buffer, state=state, action=action, reward=reward, args=arg_dict)
# add phase to sim manager
manager.add_phase("manipulation", manipulation, start=True)
# Run the sim
if runtype == 'run':
for k in range(int(args['epochs']/len(x))):
manager.run()
manager.phase_manager.phase_dict['manipulation'].reset()
'''
if k % args['evaluate'] == 0:
manager.evaluate()
manager.phase_manager.phase_dict['manipulation'].reset()
# manager.train()
'''
manager.save_network(args['save_path']+'policy')
# replay_buffer.save_sampling('/home/orochi/mojo/mojo-grasp/demos/rl_demo/data/hard_random_sampling/sampling')
print('training done, creating episode_all')
d = data_processor(args['save_path']+'Train/')
d.load_data()
d.save_all()
manager.phase_manager.phase_dict['manipulation'].controller.policy.save_sampling()
elif runtype == 'eval':
manipulation.load_policy(args['save_path']+'policy')
manager.evaluate()
manager.phase_manager.phase_dict['manipulation'].reset()
elif runtype == 'cont':
manipulation.load_policy(args['save_path']+'policy')
for k in range(int(args['epochs']/len(x))):
manager.run()
manager.phase_manager.phase_dict['manipulation'].reset()
manager.save_network(args['save_path']+'policy_2')
print('training done, creating episode_all')
d = data_processor(args['save_path'])
d.load_data()
d.save_all()
elif runtype == 'transfer':
manipulation.load_policy(args['load_path']+'policy')
manager.evaluate()
manager.phase_manager.phase_dict['manipulation'].reset()
for k in range(int(args['epochs']/len(x))):
manager.run()
manager.phase_manager.phase_dict['manipulation'].reset()
if k % 10 == 9:
manager.evaluate()
manager.phase_manager.phase_dict['manipulation'].reset()
def main(run_id):
print(run_id)
folder_names = ['0_ftp_wja_tt','1_ftp_wja_ft','2_ftp_n_tt','3_ftp_n_ft','4_ja_wja_tt',
'5_ja_wja_ft','6_ja_n_tt','7_ja_n_ft']
overall_path = pathlib.Path(__file__).parent.resolve()
run_path = overall_path.joinpath('demos/rl_demo/data/hpc_2-5')
final_path = run_path.joinpath(folder_names[run_id-1])
print(str(final_path))
run_pybullet(str(final_path) + '/experiment_config.json',runtype='run')
# run_pybullet('/home/orochi/mojo/mojo-grasp/demos/rl_demo/data/6_ik_kegan_point_split/experiment_config.json',runtype='eval')
# run_pybullet('/home/orochi/mojo/mojo-grasp/demos/rl_demo/data/a_throwaway/experiment_config.json',runtype='run')
print('buttz')
if __name__ == '__main__':
# print(sys.argv)
main(int(sys.argv[1]))