File size: 4,443 Bytes
f96995c |
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 |
from pathlib import Path
import random
from tqdm import tqdm, trange
import hydra
from omegaconf import DictConfig, OmegaConf
import numpy as np
import torch
import torch.nn as nn
import torch.backends.cudnn
import math
import pyvista as pv
from pgnd.utils import get_root, mkdir
from pgnd.ffmpeg import make_video
from train.pv_utils import Xvfb, get_camera_custom
@torch.no_grad()
def render(
cfg,
log_root,
iteration,
episode_names,
eval_dirname='eval',
eval_postfix='',
dataset_name='',
start_step=None,
end_step=None,
clean_bg=False,
):
clean_bg = False
if dataset_name == '':
eval_name = f'{cfg.train.name}/{eval_dirname}/{iteration:06d}'
else:
eval_name = f'{cfg.train.name}/{eval_dirname}/{dataset_name}/{iteration:06d}'
render_type = 'pv'
exp_root: Path = log_root / eval_name
state_root: Path = exp_root / 'state'
image_root: Path = exp_root / render_type
mkdir(image_root, overwrite=cfg.overwrite, resume=cfg.resume)
video_path_list = []
for episode_idx, episode in enumerate(episode_names):
plotter = pv.Plotter(lighting='three lights', off_screen=True, window_size=(cfg.render.width, cfg.render.height))
plotter.set_background('white')
plotter.camera_position = get_camera_custom(cfg.render.center, cfg.render.distance, cfg.render.azimuth, cfg.render.elevation)
plotter.enable_shadows()
# add bounding box
scale_x = cfg.sim.num_grids[0] / (cfg.sim.num_grids[0] - 2 * cfg.render.bound)
scale_y = cfg.sim.num_grids[1] / (cfg.sim.num_grids[1] - 2 * cfg.render.bound)
scale_z = cfg.sim.num_grids[2] / (cfg.sim.num_grids[2] - 2 * cfg.render.bound)
scale = np.array([scale_x, scale_y, scale_z])
scale_mean = np.power(np.prod(scale), 1 / 3)
bbox = pv.Box(bounds=[0, 1, 0, 1, 0, 1])
if not clean_bg:
plotter.add_mesh(bbox, style='wireframe', color='black')
# add axis
if not clean_bg:
for axis, color in enumerate(['r', 'g', 'b']):
mesh = pv.Arrow(start=[0, 0, 0], direction=np.eye(3)[axis], scale=0.2)
plotter.add_mesh(mesh, color=color, show_scalar_bar=False)
episode_state_root = state_root / episode
episode_image_root = image_root / episode
mkdir(episode_image_root, overwrite=cfg.overwrite, resume=cfg.resume)
ckpt_paths = list(sorted(episode_state_root.glob('*.pt'), key=lambda x: int(x.stem)))
for i, path in enumerate(tqdm(ckpt_paths, desc=render_type)):
if i % cfg.render.skip_frame != 0:
continue
ckpt = torch.load(path, map_location='cpu')
p_x = ckpt['x'].cpu().detach().numpy()
x = (p_x - 0.5) * scale + 0.5
grippers = ckpt['grippers'].cpu().detach().numpy()
n_eef = grippers.shape[0]
radius = 0.5 * np.power((0.5 ** 3) / x.shape[0], 1 / 3) * scale_mean
x = np.clip(x, radius, 1 - radius)
polydata = pv.PolyData(x)
plotter.add_mesh(polydata, style='points', name='object', render_points_as_spheres=True, point_size=radius * cfg.render.radius_scale, color=list(cfg.render.reflectance))
for j in range(n_eef):
gripper = pv.Sphere(center=grippers[j, :3], radius=0.04)
plotter.add_mesh(gripper, color='blue', name=f'gripper_{j}')
if 'gripper_x' in ckpt:
gripper_points = ckpt['gripper_x'].cpu().detach().numpy()
gripper_points = (gripper_points - 0.5) * scale + 0.5
gripper_points = np.clip(gripper_points, radius, 1 - radius)
gripper_polydata = pv.PolyData(gripper_points)
plotter.add_mesh(gripper_polydata, style='points', name=f'gripper_points', render_points_as_spheres=True, point_size=radius * cfg.render.radius_scale, color='blue')
plotter.show(auto_close=False, screenshot=str(episode_image_root / f'{i // cfg.render.skip_frame:04d}.png'))
plotter.close()
make_video(episode_image_root, image_root / f'{episode}{eval_postfix}.mp4', '%04d.png', cfg.render.fps)
video_path_list.append(image_root / f'{episode}{eval_postfix}.mp4')
return video_path_list
@torch.no_grad()
def do_train_pv(*args, **kwargs):
with Xvfb():
ret = render(*args, **kwargs)
return ret
|