109 lines
3.6 KiB
Python
109 lines
3.6 KiB
Python
import itertools
|
|
import numpy as np
|
|
import rospy
|
|
|
|
from .policy import BasePolicy
|
|
from vgn.utils import look_at, spherical_to_cartesian
|
|
|
|
|
|
class NextBestView(BasePolicy):
|
|
def __init__(self, rate, filter_grasps):
|
|
super().__init__(rate, filter_grasps)
|
|
|
|
def activate(self, bbox):
|
|
super().activate(bbox)
|
|
|
|
def update(self, img, extrinsic):
|
|
# Integrate latest measurement
|
|
self.integrate_img(img, extrinsic)
|
|
|
|
# Generate viewpoints
|
|
views = self.generate_viewpoints()
|
|
|
|
# Evaluate viewpoints
|
|
gains = [self.compute_ig(v) for v in views]
|
|
costs = [self.compute_cost(v) for v in views]
|
|
utilities = gains / np.sum(gains) - costs / np.sum(costs)
|
|
|
|
# Visualize
|
|
self.visualizer.views(self.base_frame, self.intrinsic, views, utilities)
|
|
|
|
# Determine next-best-view
|
|
nbv = views[np.argmax(utilities)]
|
|
|
|
if self.check_done():
|
|
self.best_grasp = self.compute_best_grasp()
|
|
self.done = True
|
|
return
|
|
|
|
return nbv.inv() # the controller expects T_cam_base
|
|
|
|
def generate_viewpoints(self):
|
|
r, h = 0.14, 0.2
|
|
thetas = np.arange(1, 4) * np.deg2rad(30)
|
|
phis = np.arange(1, 6) * np.deg2rad(60)
|
|
views = []
|
|
for theta, phi in itertools.product(thetas, phis):
|
|
eye = self.center + np.r_[0, 0, h] + spherical_to_cartesian(r, theta, phi)
|
|
target = self.center
|
|
up = np.r_[1.0, 0.0, 0.0]
|
|
views.append(look_at(eye, target, up).inv())
|
|
return views
|
|
|
|
def compute_ig(self, view, downsample=20):
|
|
fx = self.intrinsic.fx / downsample
|
|
fy = self.intrinsic.fy / downsample
|
|
cx = self.intrinsic.cx / downsample
|
|
cy = self.intrinsic.cy / downsample
|
|
|
|
T_cam_base = view.inv()
|
|
corners = np.array([T_cam_base.apply(p) for p in self.bbox.corners]).T
|
|
u = (fx * corners[0] / corners[2] + cx).round().astype(int)
|
|
v = (fy * corners[1] / corners[2] + cy).round().astype(int)
|
|
u_min, u_max = u.min(), u.max()
|
|
v_min, v_max = v.min(), v.max()
|
|
|
|
t_min = 0.2
|
|
t_max = corners[2].max() # TODO This bound might be a bit too short
|
|
t_step = 0.01
|
|
|
|
view = self.T_task_base * view # We'll work in the task frame from now on
|
|
|
|
tsdf_grid, voxel_size = self.tsdf.get_grid(), self.tsdf.voxel_size
|
|
|
|
def get_voxel_at(p):
|
|
index = (p / voxel_size).astype(int)
|
|
return index if (index >= 0).all() and (index < 40).all() else None
|
|
|
|
for u in range(u_min, u_max):
|
|
for v in range(v_min, v_max):
|
|
origin = view.translation
|
|
direction = np.r_[(u - cx) / fx, (v - cy) / fy, 1.0]
|
|
direction = view.rotation.apply(direction / np.linalg.norm(direction))
|
|
|
|
self.visualizer.rays(self.task_frame, origin, [direction])
|
|
rospy.sleep(0.1)
|
|
|
|
t, tsdf_prev = t_min, -1.0
|
|
while t < t_max:
|
|
p = origin + t * direction
|
|
t += t_step
|
|
|
|
self.visualizer.point(self.task_frame, p)
|
|
rospy.sleep(0.1)
|
|
|
|
index = get_voxel_at(p)
|
|
if index is not None:
|
|
i, j, k = index
|
|
tsdf = -1 + 2 * tsdf_grid[i, j, k] # Open3D maps tsdf to [0,1]
|
|
if tsdf * tsdf_prev < 0 and tsdf_prev > -1: # Crossed a surface
|
|
break
|
|
tsdf_prev = tsdf
|
|
return 1.0
|
|
|
|
def compute_cost(self, view):
|
|
return 1.0
|
|
|
|
def check_done(self):
|
|
return len(self.viewpoints) == 4
|