-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_dp_agent_zmq.py
123 lines (106 loc) · 3.59 KB
/
test_dp_agent_zmq.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
import argparse
import os
import pickle
import time
import numpy as np
import torch
from agents.dp_agent_zmq import BimanualDPAgent
from learning.dp.data_processing import iterate
torch.cuda.set_device(0)
def main(args):
hand_uppers = np.array([110.0, 110.0, 110.0, 110.0, 90.0, 120.0])
hand_lowers = np.array([5.0, 5.0, 5.0, 5.0, 5.0, 5.0])
data = iterate(args.data_dir)
num_diffusion_iters = args.num_diffusion_iters
dp_agent = BimanualDPAgent(
ckpt_path="best.ckpt",
host="localhost",
port="4321",
)
dp_agent.compile_inference(data[0], num_diffusion_iters=num_diffusion_iters)
controls = []
pred_actions = []
delta_action = []
last_action = data[0]["control"]
# start = time.time()
# end = time.time()
# ctime = end - start
# print(f"compilation time: {ctime}")
start = time.time()
max_infer_time = 0
for i, obs in enumerate(data):
time.sleep(0.1)
control = obs["control"]
delta_action.append(control - last_action)
last_action = control
controls.append(control)
if i != 0:
obs["joint_positions"][list(range(6)) + list(range(12, 18))] = pred_actions[
-1
][list(range(6)) + list(range(12, 18))]
obs["joint_positions"][6:12] = (
pred_actions[-1][6:12] * (hand_uppers - hand_lowers) + hand_lowers
)
obs["joint_positions"][18:24] = (
pred_actions[-1][18:24] * (hand_uppers - hand_lowers) + hand_lowers
)
infer_start = time.time()
pred_actions.append(dp_agent.act(obs))
infer_time = time.time() - infer_start
max_infer_time = max(max_infer_time, infer_time)
print("num_diffusion_iters:", num_diffusion_iters)
print("time:", time.time() - start)
print("Hz:", len(data) / (time.time() - start))
print("max_infer_time:", max_infer_time)
print("lowest_freq:", 1 / max_infer_time)
pred_actions = np.array(pred_actions)
controls = np.array(controls)
mse = np.mean(np.abs((pred_actions - controls)), axis=0)
mean_delta_action = np.mean(np.abs(delta_action), axis=0)
print_str = "\n".join(
[
"mse:",
str(mse.tolist()),
"\n",
"mean_delta_action:",
str(mean_delta_action.tolist()),
"\nfinal_diff:",
str((pred_actions[-1] - controls[-1]).tolist()),
]
)
print_str += "\n"
print_str += (
"mse: "
+ str(mse.mean())
+ " mean_delta_action: "
+ str(mean_delta_action.mean())
)
print(print_str)
# save print_str as txt in ckpt dir
ckpt_dir = os.path.dirname(args.ckpt_path)
with open(
os.path.join(ckpt_dir, f"eval_stats_{num_diffusion_iters}.txt"), "w"
) as f:
f.write(print_str)
traj_name = os.path.basename(args.data_dir)
save_path = os.path.join(ckpt_dir, f"openloop_{traj_name}_{num_diffusion_iters}")
os.makedirs(save_path, exist_ok=True)
for i in range(len(pred_actions)):
with open(os.path.join(save_path, str(i) + ".pkl"), "wb") as f:
pickle.dump(
{
"control": pred_actions[i],
"joint_positions": data[i]["joint_positions"],
},
f,
)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--data_dir",
type=str,
default="data/",
)
parser.add_argument("--num_diffusion_iters", default=15, type=int)
args = parser.parse_args()
main(args)