def run():
app = QtWidgets.QApplication(sys.argv)
form = Form()
ui = Ui_Form()
ui.setupUi(form)
ui.set_controller(controller.make_controller(ui))
form.show()
sys.exit(app.exec_())
def set_controller(self, arg, clear=True):
if type(arg) is str:
cont = load_controller(arg)
else:
name = arg.pop("name")
cont = make_controller(name, arg)
if not clear:
self.controllers.append(cont)
else:
self.controllers = [cont]
'Y3': Axis(bpr, 'BRICK1CS3:Y', 8, 3)
}
brick_groups = {
'g1': '1,2->A,B',
'g2': '3,4->I',
'g3': 'MIXED CS3',
'g4': 'MIXED CS2'
}
brick_cs = {
'cs2': CoordSys('BRICK1:CS2', 2, 'CS2'),
'cs3': CoordSys('BRICK1:CS3', 3, 'CS3')
}
MyBrick = make_controller(brick_axes, brick_cs, brick_groups, brick_pv_root)
class TestBrick(MyBrick):
# noinspection PyAttributeOutsideInit
def startup(self, init=True):
a = self.axes
r = self.real_axes
# axis aliases
self.height = self.X3
self.angle = self.Y3
self.jack1 = self.m3
self.jack2 = self.m4
# useful aliases for beamline test authoring
def initialize_settings(sigma_init=0.1, sigma_decay=0.9999):
global population, filebase, game, controller, num_params, es, PRECISION, SOLUTION_PACKET_SIZE, RESULT_PACKET_SIZE
population = num_worker * num_worker_trial
filedir = 'results/{}/{}/log/'.format(exp_name, env_name)
if not os.path.exists(filedir):
os.makedirs(filedir)
filebase = filedir + env_name + '.' + optimizer + '.' + str(
num_episode) + '.' + str(population)
controller = make_controller(args=config_args)
num_params = controller.param_count
print("size of model", num_params)
if optimizer == 'ses':
ses = PEPG(num_params,
sigma_init=sigma_init,
sigma_decay=sigma_decay,
sigma_alpha=0.2,
sigma_limit=0.02,
elite_ratio=0.1,
weight_decay=0.005,
popsize=population)
es = ses
elif optimizer == 'ga':
ga = SimpleGA(num_params,
sigma_init=sigma_init,
sigma_decay=sigma_decay,
sigma_limit=0.02,
elite_ratio=0.1,
weight_decay=0.005,
popsize=population)
es = ga
elif optimizer == 'cma':
cma = CMAES(num_params, sigma_init=sigma_init, popsize=population)
es = cma
elif optimizer == 'pepg':
pepg = PEPG(num_params,
sigma_init=sigma_init,
sigma_decay=sigma_decay,
sigma_alpha=0.20,
sigma_limit=0.02,
learning_rate=0.01,
learning_rate_decay=1.0,
learning_rate_limit=0.01,
weight_decay=0.005,
popsize=population)
es = pepg
else:
oes = OpenES(num_params,
sigma_init=sigma_init,
sigma_decay=sigma_decay,
sigma_limit=0.02,
learning_rate=0.01,
learning_rate_decay=1.0,
learning_rate_limit=0.01,
antithetic=antithetic,
weight_decay=0.005,
popsize=population)
es = oes
PRECISION = 10000
SOLUTION_PACKET_SIZE = (5 + num_params) * num_worker_trial
RESULT_PACKET_SIZE = 4 * num_worker_trial
import numpy as np
import random
import os
import gym
from env import make_env
from controller import make_controller
from utils import PARSER
args = PARSER.parse_args()
dir_name = 'results/{}/{}/record'.format(args.exp_name, args.env_name)
if not os.path.exists(dir_name):
os.makedirs(dir_name)
controller = make_controller(args=args)
total_frames = 0
env = make_env(args=args,
render_mode=args.render_mode,
full_episode=args.full_episode,
with_obs=True,
load_model=False)
for trial in range(args.max_trials):
try:
random_generated_int = random.randint(0, 2**31 - 1)
filename = dir_name + "/" + str(random_generated_int) + ".npz"
recording_frame = []
recording_action = []
recording_reward = []
def main():
print("Setting niceness to 19")
if "nice" in os.__dict__:
os.nice(19)
args = PARSER.parse_args()
def make_env_with_args():
return make_env(args=args, keep_image=True, wrap_rnn=False)
dir_name = get_path(args, "record", create=True)
controller = None
if args.extract_use_controller:
controller = make_controller(args=args)
env = make_env_with_args()
has_camera_data = isinstance(
env.observation_space,
gym.spaces.Dict) and "camera" in env.observation_space.spaces
format_str = "[{success:s}] {done:s} after {frames:4d} frames, reward {reward:6.1f} " \
"(Total: {total_frames:7d} frames, {successful_trials:3d}/{total_trials:3d} successful trials)"
total_frames = 0
successful_trials = 0
for trial in range(args.max_trials):
try:
seed = random.randint(0, 2**31 - 1)
filename = dir_name / (str(seed) + ".npz")
np.random.seed(seed)
env.seed(seed)
recording_image = []
recording_camera = []
recording_action = []
recording_reward = []
recording_done = []
# random policy
if args.extract_use_controller:
controller.init_random_model_params(stddev=np.random.rand() *
0.01)
repeat_action = np.random.randint(1, 11)
action = [0] * args.a_width
total_reward = 0
obs = env.reset()
frame = 0
ended_early = False
for frame in range(args.max_frames):
# Save current observation
recording_image.append(obs["image"])
if has_camera_data:
recording_camera.append(obs["camera"])
# Get next action (random)
if not args.extract_repeat_actions or frame % repeat_action == 0:
if args.extract_use_controller:
action = controller.get_action(obs["features"])
else:
action = np.random.rand(args.a_width) * 2.0 - 1.0
if args.extract_repeat_actions:
repeat_action = np.random.randint(1, 11)
# Save action
recording_action.append(action)
# Perform action
obs, reward, done, _info = env.step(action)
total_reward += reward
# Save reward and done flag
recording_reward.append(reward)
recording_done.append(done)
# Stop when done
if done:
ended_early = True
break
total_frames += (frame + 1)
enough_frames = len(recording_image) >= args.min_frames
# Save episode to disk (if it has required minimum length)
if enough_frames:
successful_trials += 1
recording_image = np.array(recording_image, dtype=np.uint8)
recording_camera = np.array(recording_camera, dtype=np.float16)
recording_action = np.array(recording_action, dtype=np.float16)
recording_reward = np.array(recording_reward, dtype=np.float16)
recording_done = np.array(recording_done, dtype=np.bool)
data = {
"image": recording_image,
"action": recording_action,
"reward": recording_reward,
"done": recording_done
}
if has_camera_data:
data["camera"] = recording_camera
np.savez_compressed(str(filename), **data)
print(
format_str.format(success="O" if enough_frames else " ",
done="Done" if ended_early else "Stop",
frames=frame + 1,
reward=total_reward,
total_frames=total_frames,
successful_trials=successful_trials,
total_trials=trial + 1))
except gym.error.Error as e:
print("Gym raised an error: " + str(e))
env.close()
env = make_env_with_args()
env.close()