def get_farthest_in_range(self, start_point: Matrix,
path: List[Matrix]) -> Tuple[int, Matrix]:
farthest = 0
for i, point in enumerate(path):
d = dist(tf.translation_from_matrix(point),
tf.translation_from_matrix(start_point))
if d > self.max_stride:
break
else:
farthest = i
return farthest, path[farthest].copy()
def interpolate(start: np.ndarray, stop: np.ndarray, phase: float) -> np.ndarray:
start_trans = tf.translation_from_matrix(start)
stop_trans = tf.translation_from_matrix(stop)
shift_trans = phase * (stop_trans - start_trans)
inter_trans = start_trans + shift_trans
trans_matrix = tf.translation_matrix(inter_trans)
start_rot = np.array(tf.euler_from_matrix(start, axes='sxyz'))
stop_rot = np.array(tf.euler_from_matrix(stop, axes='sxyz'))
shift_rot = phase * (stop_rot - start_rot)
inter_rot = start_rot + shift_rot
rot_matrix = tf.euler_matrix(*inter_rot, axes='sxyz')
result = tf.concatenate_matrices(trans_matrix, rot_matrix)
return result
def updateData(self, feature, prop):
r"""If a property of the handled feature has changed,
we have the chance to handle this here
See Also
--------
https://www.freecadweb.org/wiki/Scripted_objects
"""
debug("ViewProviderAnchor/updateData")
p = feature.getPropertyByName("p")
u = feature.getPropertyByName("u")
v = feature.getPropertyByName("v")
self.transform.translation.setValue((p[0], p[1], p[2]))
at = anchor_transformation(p0=(0, 0, 0),
u0=(0, -1, 0),
v0=(0, 0, 1),
p1=(p[0], p[1], p[2]),
u1=(u[0], u[1], u[2]),
v1=(v[0], v[1], v[2]))
t = translation_from_matrix(at)
self.transform.translation.setValue((t[0], t[1], t[2]))
angle, direction, point = rotation_from_matrix(at)
# print("angle : %f" % angle)
# print("direction : %s" % str(direction))
# print("point : %s" % str(point))
self.transform.rotation.setValue(coin.SbVec3f(direction), angle)
def determine_obj_frame(self): if self.detected_artags != None: eng_poses = [] ar_tfs = [] for da in self.detected_artags.markers: ar_pose = th.aruco_to_pose_stamp(da) eng_pose = self.get_engine_pose_from_ps(ar_pose) eng_poses.append(eng_pose) ar_tfs.append(th.pose_stamp_to_tf(eng_pose)) ar_tf_sum = np.zeros((4,4)) for tf in ar_tfs: ar_tf_sum = ar_tf_sum + tf/len(ar_tfs) engine_frame_combined = ar_tf_sum tran = transformations.translation_from_matrix(engine_frame_combined), eul = transformations.euler_from_matrix(engine_frame_combined) self.updated_engine_pose = str(tran[0][0]) + ' ' + str(tran[0][1]) + ' ' + str(tran[0][2]) + ' ' + str(eul[0]) + ' ' + str(eul[1]) + ' ' + str(eul[2]) ps = th.tf_to_pose_stamp(engine_frame_combined, 'engine_frame_combined') th.broadcast_transform(ps, global_frame) return engine_frame_combined else: return None
def back_front_thread():
global published_transform
rate = rospy.Rate(10)
while True:
time = rospy.Time.now()
num = 0
sum_mat = None
#print("Time diff = %s" %(str(time.to_sec() - front_transform["time"].to_sec()) if front_transform["time"] is not None else "??"))
if front_transform["transform"] is not None and time.to_sec(
) - front_transform["time"].to_sec() < 1:
sum_mat = front_transform["transform"]
num = 1.0
# BRS Let's not use averages - seems to cause normailization errors?
if back_transform["transform"] is not None and time.to_sec(
) - back_transform["time"].to_sec() < 1:
sum_mat = back_transform[
"transform"] if sum_mat is None else sum_mat + back_transform[
"transform"]
num = num + 1.0
if num > 0:
transform = sum_mat / num
published_transform = transform
trans = tr.translation_from_matrix(transform)
rot = tr.quaternion_from_matrix(transform)
r, p, y = tr.euler_from_quaternion(rot)
rot = tr.quaternion_from_euler(r, p, y)
br.sendTransform(trans, rot, time, "odom", "map")
# elif published_transform is not None:
# transform = published_transform
# trans = tr.translation_from_matrix(transform)
# rot = tr.quaternion_from_matrix(transform)
# br.sendTransform(trans, rot, time, "odom", "map")
rate.sleep()
def inframe(self, pose, frame):
""" Transform a pose from one frame to another one.
Uses transformation matrices. Could be refactored to use directly
quaternions.
"""
pose = self.get(pose)
if pose['frame'] == "map":
orig = numpy.identity(4)
else:
orig = self._to_mat4(self.get(pose["frame"]))
if frame == "map":
dest = numpy.identity(4)
else:
dest = numpy.linalg.inv(self._to_mat4(self.get(frame)))
poseMatrix = self._to_mat4(pose)
transf = numpy.dot(dest, orig)
transformedPose = numpy.dot(transf, poseMatrix)
qx,qy,qz,qw = transformations.quaternion_from_matrix(transformedPose)
x,y,z = transformations.translation_from_matrix(transformedPose)
return {"x":x,
"y":y,
"z":z,
"qx":qx,
"qy":qy,
"qz":qz,
"qw":qw,
"frame": frame}
def draw(self, surface, baseTransform, parameters):
armLength = tr.translation_matrix([self.length, 0, 0])
rPitch = tr.rotation_matrix(parameters[0], yaxis)
rYaw = tr.rotation_matrix(parameters[1], zaxis)
transform = tr.concatenate_matrices(baseTransform, rYaw, rPitch,
armLength)
start = tr.translation_from_matrix(baseTransform)
end = tr.translation_from_matrix(transform)
# Only use the y and z coords for a quick and dirty orthogonal projection
pygame.draw.line(surface, (0, 0, 0), start[::2], end[::2], 5)
# Draw angle constraints
#pygame.draw.aaline(surface, (100, 0, 0), position, position + to_vector(self.pitch.absolute_min(base_angle), 30))
#pygame.draw.aaline(surface, (100, 0, 0), position, position + to_vector(self.pitch.absolute_max(base_angle), 30))
if self.after is not None:
self.after.draw(surface, transform, parameters[2:])
def error(self, target_pos, baseTransform, parameters):
transform = np.dot(baseTransform,
self.transform(parameters[0], parameters[1]))
if self.after is None:
end = tr.translation_from_matrix(transform)
return distance(target_pos, end)
else:
return self.after.error(target_pos, transform, parameters[2:])
def _joints_impl(self, baseTransform, parameters):
transform = np.dot(baseTransform,
self.transform(parameters[0], parameters[1]))
end = tr.translation_from_matrix(transform)
if self.after is not None:
return [end] + self.after._joints_impl(transform, parameters[2:])
else:
return [end]
def tf_to_dof(T):
if T.shape == (12, ):
T.shape = (3, 4)
if T.shape == (3, 4):
T = np.concatenate((T, np.array([[0.0, 0.0, 0.0, 1.0]])), axis=0)
assert T.shape == (4, 4), 'T.shape is {}, not (4,4)'.format(T.shape)
translation = tr.translation_from_matrix(T)
# euler_angles = np.array(tr.euler_from_matrix(T[0:3, 0:3]))
euler_angles = np.array(tr.euler_from_matrix(T))
return np.concatenate((translation, euler_angles), axis=0)
def tf_to_quat(T):
if T.shape == (12, ):
T.shape = (3, 4)
if T.shape == (3, 4):
T = np.concatenate((T, np.array([[0.0, 0.0, 0.0, 1.0]])), axis=0)
assert T.shape == (4, 4)
translation = tr.translation_from_matrix(T)
# quaternions = np.array(tr.quaternion_from_matrix(T[0:3, 0:3]))
quaternions = np.array(tr.quaternion_from_matrix(T))
return np.concatenate((translation, quaternions), axis=0)
def error(self, target_pos, baseTransform, parameters):
armLength = tr.translation_matrix([self.length, 0, 0])
rPitch = tr.rotation_matrix(parameters[0], yaxis)
rYaw = tr.rotation_matrix(parameters[1], zaxis)
transform = tr.concatenate_matrices(baseTransform, rYaw, rPitch,
armLength)
if self.after is None:
end = tr.translation_from_matrix(transform)
return distance(target_pos, end)
else:
return self.after.error(target_pos, transform, parameters[2:])
def generate_leg(leg: str, params: list):
front_signs = {'front': '', 'rear': '-'}
side_signs = {'left': '', 'right': '-'}
lower_limits = {'left': '0', 'right': str(-pi)}
upper_limits = {'left': str(pi), 'right': '0'}
with open('../urdf/leg.urdf.template', 'r') as file:
leg_template = Template(file.read())
mappings = {'l_width': 0.02, 'leg': leg}
front, side = leg.split('_')
mappings['front_sign'] = front_signs[front]
mappings['side_sign'] = side_signs[side]
mappings['lower_limit'] = lower_limits[side]
mappings['upper_limit'] = upper_limits[side]
for param in params:
try:
d = float(param['d'])
except ValueError:
d = 0.0
try:
th = float(param['th'])
except ValueError:
th = 0.0
try:
a = float(param['a'])
except ValueError:
a = 0.0
try:
al = float(param['al'])
except ValueError:
al = 0.0
tz = translation_matrix((0, 0, d))
rz = rotation_matrix(th, zaxis)
tx = translation_matrix((a, 0, 0))
rx = rotation_matrix(al, xaxis)
matrix = concatenate_matrices(tz, rz, tx, rx)
rpy = euler_from_matrix(matrix)
xyz = translation_from_matrix(matrix)
name = param['joint']
mappings[f'{name}_j_xyz'] = '{} {} {}'.format(*xyz)
mappings[f'{name}_j_rpy'] = '{} {} {}'.format(*rpy)
mappings[f'{name}_l_xyz'] = '{} 0 0'.format(xyz[0] / 2.)
mappings[f'{name}_l_rpy'] = '0 0 0'
mappings[f'{name}_l_len'] = str(a)
return leg_template.substitute(mappings)
def build_joint(index, parent_index):
matrix = model['bone_original_matrix'][index]
x, y, z = tf.translation_from_matrix(matrix.T)
bone_pool.append(
pmx.Bone(name=model['bone_name'][index],
english_name=model['bone_name'][index],
position=common.Vector3(-x, y, -z),
parent_index=parent_index,
layer=0,
flag=0))
bone_pool[-1].setFlag(pmx.BONEFLAG_CAN_ROTATE, True)
bone_pool[-1].setFlag(pmx.BONEFLAG_IS_VISIBLE, True)
bone_pool[-1].setFlag(pmx.BONEFLAG_CAN_MANIPULATE, True)
def matrix_to_pose(matrix):
pose = Pose()
trans = tf.translation_from_matrix(matrix)
quat = tf.quaternion_from_matrix(matrix)
pose.position.x = trans[0]
pose.position.y = trans[1]
pose.position.z = trans[2]
pose.orientation.w = quat[0]
pose.orientation.x = quat[1]
pose.orientation.y = quat[2]
pose.orientation.z = quat[3]
return pose
def errors(reference, target):
if len(reference) != len(target):
raise ValueError("reference and target should be of same length! len(reference): %d, len(target): %d" % (len(reference), len(target)) )
F_ref = to_homogeneous(reference)
F_tar = to_homogeneous(target)
F_ref_inv = inverse_matrix(F_ref)
T_tar_ref = concatenate_matrices(F_ref_inv, F_tar)
trans = translation_from_matrix(T_tar_ref)
ang, foo, bar = rotation_from_matrix(T_tar_ref)
return (numpy.sum( numpy.square(trans) ), ang*ang)
def scale_and_translate(Tlistgroundtruth, Tlistartoolkit):
ground_truth_pointcloud = np.asarray([utils.apply_transform(T, np.zeros(3))
for T in Tlistgroundtruth])
artoolkit_pointcloud = np.asarray([utils.apply_transform(T, np.zeros(3))
for T in Tlistartoolkit])
Tscale = tf.affine_matrix_from_points(artoolkit_pointcloud.T,
ground_truth_pointcloud.T,
shear=False,
scale=True)
#print("Scaling by {0}".format(tf.scale_from_matrix(Tscale)[0]))
try:
print("translation by {0}".format(tf.translation_from_matrix(Tscale)))
print("rotation by {0}".format(tf.rotation_from_matrix(Tscale)[0]))
except ValueError:
pass
Tlistartoolkit = [tf.concatenate_matrices(Tscale, T) for T in Tlistartoolkit]
return Tlistartoolkit
def publish_transform(src: str, dst: str, pose: np.ndarray):
trans = tf.translation_from_matrix(pose)
rot = tf.quaternion_from_matrix(pose)
transform = TransformStamped()
transform.header.stamp = header_stamp(clock.now())
transform.header.frame_id = src
transform.child_frame_id = dst
transform.transform.translation.x = trans[0]
transform.transform.translation.y = trans[1]
transform.transform.translation.z = trans[2]
transform.transform.rotation.w = rot[0]
transform.transform.rotation.x = rot[1]
transform.transform.rotation.y = rot[2]
transform.transform.rotation.z = rot[3]
msg = TFMessage(transforms=[transform])
broadcaster.publish(msg)
def from_homogeneous(M,n):
if n != 7 and n != 3:
raise ValueError("Only poses with 3 or 7 elements supported! this one has %d" % n)
trans = translation_from_matrix(M)
if n == 7:
quat = quaternion_from_matrix(M)
out = list(trans)
out.extend(quaternion_imag(quat))
out.append(quaternion_real(quat))
else:
ang, foo, bar = rotation_from_matrix(M)
out = list(trans[0:2])
out.append(ang)
if len(out) != n:
raise ValueError("Wrong output length: %d should be %d" %(len(out), n))
return out
def inframe(self, pose, frame):
""" Transform a pose from one frame to another one.
Uses transformation matrices. Could be refactored to use directly
quaternions.
"""
pose = self.get(pose)
if pose["frame"] == frame:
return pose
if pose['frame'] == "map":
orig = numpy.identity(4)
else:
orig = self._to_mat4(self.get(pose["frame"]))
if frame == "map":
dest = numpy.identity(4)
else:
dest = numpy.linalg.inv(self._to_mat4(self.get(frame)))
pose_matrix = self._to_mat4(pose)
transf = numpy.dot(dest, orig)
transformedPose = numpy.dot(transf, pose_matrix)
qx, qy, qz, qw = transformations.quaternion_from_matrix(
transformedPose)
x, y, z = transformations.translation_from_matrix(transformedPose)
return {
"x": float(x),
"y": float(y),
"z": float(z),
"qx": float(qx),
"qy": float(qy),
"qz": float(qz),
"qw": float(qw),
"frame": frame
}
def calc_legs(self, pose: ndarray, base_pose: ndarray,
left_side: bool) -> Dict[str, float]:
if len(self.base_to_hip) == 0:
self.init_transforms()
names = []
positions = []
legs_positions = self.triangle(pose, left_side)
legs = ['l1', 'r2', 'l3'] if left_side else ['r1', 'l2', 'r3']
for leg, position in zip(legs, legs_positions):
base_to_hip = self.base_to_hip[leg]
relative = tf.concatenate_matrices(tf.inverse_matrix(base_to_hip),
tf.inverse_matrix(base_pose),
position)
point = tf.translation_from_matrix(relative)
a, b, g = self.inverse_kinematics(*point)
positions += [a, b, g]
names += [
f'coxa_joint_{leg}', f'femur_joint_{leg}', f'tibia_joint_{leg}'
]
return {name: position for name, position in zip(names, positions)}
def process_markers(markers):
global listener, br, marker_trans_mat
sum_mat = None
num = 0
t_latest = rospy.Time(0)
for marker in markers.markers:
print("Seeing marker %s" % marker.id)
marker_link = '/Marker' + str(marker.id) + "__link"
marker_id = "/Marker" + str(marker.id)
tw = listener.getLatestCommonTime(marker_link, 'world')
to = listener.getLatestCommonTime(marker_id, 'odom')
t = tw if tw > to else to
t_latest = t if t > t_latest else t_latest
(trans_w, rot_w) = listener.lookupTransform(marker_link, 'world', tw)
(trans_o, rot_o) = listener.lookupTransform(marker_id, 'odom', to)
w_mat = numpy.dot(tr.translation_matrix(trans_w),
tr.quaternion_matrix(rot_w))
t_o = tr.concatenate_matrices(tr.translation_matrix(trans_o),
tr.quaternion_matrix(rot_o))
# Need to do the transform equivalent to 0.25 0 0 0 0.5 -0.5 -0.5 0.5 on marker_id
#t_o = numpy.dot(t_o, marker_trans_mat);
o_mat_i = tr.inverse_matrix(t_o)
mat3 = numpy.dot(w_mat, o_mat_i)
mat3 = numpy.inverse_matrix(mat3)
if sum_mat is None:
sum_mat = mat3
else:
sum_mat = sum_mat + mat3
num = num + 1
avg_mat = sum_mat / num
trans = tr.translation_from_matrix(avg_mat)
rot = tr.quaternion_from_matrix(avg_mat)
br.sendTransform(trans, rot, t_latest, "odom", "map")
def marker_hover(vehicle, marker_tracker, rs=None, hover_alt=None, debug=0):
if hover_alt is None:
hover_alt = vehicle.location.global_relative_frame.alt
# Set the PID setpoint to the desired hover altitude
# Negative because of NED coord. system
PID_Z.setpoint = -hover_alt
# Queue to contain running average for Aruco markers
pose_queue = np.zeros((RUNNING_AVG_LENGTH, 3), dtype=float)
count = 0
if debug > 0:
vehicle_pose_file = file_utils.open_file(VEHICLE_POSE_FILE)
# Hover until manual override
start_time = time.time()
time_found = time.time()
print("Tracking marker...")
while dronekit_utils.is_guided(vehicle):
# Track marker
marker_tracker.track_marker(
alt=vehicle.location.global_relative_frame.alt)
if marker_tracker.is_marker_found():
# Note the most recent time marker was found
time_found = time.time()
# Get the marker pose relative to the camera origin
marker_pose = marker_tracker.get_pose()
# Transform the marker pose into UAV body frame (NED)
marker_pose_body_ref = marker_ref_to_body_ref(
marker_pose, marker_tracker, vehicle)
# PID control; input is the UAV pose relative to the marker (hence the negatives)
control_pose = np.array([[
PID_X(-marker_pose_body_ref[0]),
PID_Y(-marker_pose_body_ref[1]),
PID_Z(-marker_pose_body_ref[2])
]])
# Keep a running average if using Aruco markers
# Yellow marker processing is too slow to use a running average
if marker_tracker.get_marker_type() == "aruco":
pose_queue[count % len(pose_queue)] = control_pose
count += 1
# Take the average
if count < RUNNING_AVG_LENGTH - 1:
pose_avg = np.sum(pose_queue, axis=0) / count
else:
pose_avg = np.average(pose_queue, axis=0)
else:
# Yellow marker
pose_avg = control_pose[0]
# Approach the marker at the current altitude until above the marker, then navigate to the hover altitude.
if np.abs(pose_avg[0]) < HOVER_THRESHOLD and np.abs(
pose_avg[1]) < HOVER_THRESHOLD:
#print("Marker Hover")
alt_hover(vehicle, pose_avg)
else:
#print("Approaching Marker")
marker_approach(vehicle, pose_avg)
if debug > 0 and rs is not None:
data = rs.read()
rs_pose = tf.quaternion_matrix([
data.rotation.w, data.rotation.x, data.rotation.y,
data.rotation.z
])
rs_pose[0][3] = data.translation.x
rs_pose[1][3] = data.translation.y
rs_pose[2][3] = data.translation.z
vehicle_pose = realsense_localization.rs_to_body(rs_pose)
vehicle_trans = np.array(
tf.translation_from_matrix(vehicle_pose))
vehicle_pose_file.write("{} {} {} {} 0 0 0 0\n".format(
time.time() - start_time, vehicle_trans[0],
vehicle_trans[1], vehicle_trans[2]))
if debug > 2:
print("Nav command: {} {} {}".format(pose_avg[0], pose_avg[1],
pose_avg[2]))
else:
if time.time() - time_found > 5:
return False
# Maintain frequency of sending commands
marker_tracker.wait()
H_body_camera[1][3] = body_offset_y
H_body_camera[2][3] = body_offset_z
H_camera_body = np.linalg.inv(H_body_camera)
H_aeroRef_aeroBody = H_body_camera.dot(H_aeroRef_aeroBody.dot(H_camera_body))
# Realign heading to face north using initial compass data
if compass_enabled == 1:
H_aeroRef_aeroBody = H_aeroRef_aeroBody.dot( tf.euler_matrix(0, 0, heading_north_yaw, 'sxyz'))
# Show debug messages here
if debug_enable == 1:
os.system('clear') # This helps in displaying the messages to be more readable
progress("DEBUG: Raw RPY[deg]: {}".format( np.array( tf.euler_from_matrix( H_T265Ref_T265body, 'sxyz')) * 180 / m.pi))
progress("DEBUG: NED RPY[deg]: {}".format( np.array( tf.euler_from_matrix( H_aeroRef_aeroBody, 'sxyz')) * 180 / m.pi))
progress("DEBUG: Raw pos xyz : {}".format( np.array( [data.translation.x, data.translation.y, data.translation.z])))
progress("DEBUG: NED pos xyz : {}".format( np.array( tf.translation_from_matrix( H_aeroRef_aeroBody))))
except Exception as e:
progress(e)
except:
send_msg_to_gcs('ERROR IN SCRIPT')
progress("Unexpected error: %s" % sys.exc_info()[0])
finally:
progress('Closing the script...')
# start a timer in case stopping everything nicely doesn't work.
signal.setitimer(signal.ITIMER_REAL, 5) # seconds...
pipe.stop()
mavlink_thread_should_exit = True
mavlink_thread.join()
def matrix_dist(a: Matrix, b: Matrix):
return dist(tf.translation_from_matrix(a), tf.translation_from_matrix(b))
def __init__(self, cfg, base_dir, sequences, frames=None):
self.cfg = cfg
self.sequences = sequences
self.curr_batch_sequence = 0
self.current_batch = 0
if (self.cfg.bidir_aug == True) and (self.cfg.batch_size % 2 != 0):
raise ValueError("Batch size must be even")
if self.cfg.data_type != "cam" and self.cfg.data_type != "lidar":
raise ValueError("lidar or camera for data type!")
frames_data_type = np.uint8
if self.cfg.data_type == "lidar":
frames_data_type = np.float16
if not frames:
frames = [None] * len(sequences)
self.input_frames = {}
self.poses = {}
# Mirror in y-z plane
self.H = np.identity(3, dtype=np.float32)
self.H[1][1] = -1.0
self.se3_ground_truth = {}
self.se3_mirror_ground_truth = {}
self.fc_ground_truth = {}
self.fc_reverse_ground_truth = {}
self.fc_mirror_ground_truth = {}
self.fc_reverse_mirror_ground_truth = {}
self.imu_measurements = {}
self.imu_measurements_mirror = {}
self.imu_measurements_reverse = {}
self.imu_measurements_reverse_mirror = {}
# This keeps track of the number of batches in each sequence
self.batch_counts = {}
# this holds the batch ids for each sequence. It is randomized and the order determines the order in
# which the batches are used
self.batch_order = {}
# this keeps track of the starting offsets in the input frames for batch id 0 in each sequence
self.batch_offsets = {}
# contains batch_cnt counters, each sequence is represented by it's index repeated for
# however many batches are in that sequence
self.sequence_ordering = None
# contains batch_size counters. Keeps track of how many batches from each sequence
# have already been used for the current epoch
self.sequence_batch = {}
self.batch_cnt = 0
self.total_frames = 0
for i_seq, seq in enumerate(sequences):
seq_loader = DataLoader(self.cfg, base_dir, seq, frames=frames[i_seq])
num_frames = seq_loader.get_num_frames()
self.input_frames[seq] = np.zeros(
[num_frames, self.cfg.input_channels, self.cfg.input_height, self.cfg.input_width],
dtype=frames_data_type)
self.poses[seq] = seq_loader.get_poses_in_corresponding_frame()
self.se3_ground_truth[seq] = np.zeros([num_frames, 7], dtype=np.float32)
self.se3_mirror_ground_truth[seq] = np.zeros([num_frames, 7], dtype=np.float32)
self.fc_ground_truth[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.fc_reverse_ground_truth[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.fc_mirror_ground_truth[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.fc_reverse_mirror_ground_truth[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.imu_measurements[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.imu_measurements_mirror[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.imu_measurements_reverse[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
self.imu_measurements_reverse_mirror[seq] = np.zeros([num_frames - 1, 6], dtype=np.float32)
for i_img in range(num_frames):
if i_img % 100 == 0:
tools.printf("Loading sequence %s %.1f%% " % (seq, (i_img / num_frames) * 100))
img = seq_loader.get_img(i_img)
self.input_frames[seq][i_img, :] = img
# now convert all the ground truth from 4x4 to xyz + quat, this is after the SE3 layer
translation = transformations.translation_from_matrix(self.poses[seq][i_img])
quat = transformations.quaternion_from_matrix(self.poses[seq][i_img])
mirror_pose = np.identity(4, dtype=np.float32)
mirror_pose[0:3, 0:3] = np.dot(self.H, np.dot(self.poses[seq][i_img][0:3, 0:3], self.H))
mirror_pose[0:3, 3] = np.dot(self.H, translation)
mirror_quat = transformations.quaternion_from_matrix(mirror_pose[0:3, 0:3])
self.se3_ground_truth[seq][i_img] = np.concatenate([translation, quat])
self.se3_mirror_ground_truth[seq][i_img] = np.concatenate([mirror_pose[0:3, 3], mirror_quat])
# relative transformation labels
if i_img + 1 < num_frames:
mirror_pose_next = np.identity(4, dtype=np.float32)
mirror_pose_next[0:3, 0:3] = np.dot(self.H, np.dot(self.poses[seq][i_img + 1][0:3, 0:3], self.H))
trans_next = transformations.translation_from_matrix(self.poses[seq][i_img + 1])
mirror_pose_next[0:3, 3] = np.dot(self.H, trans_next)
m_forward = np.dot(np.linalg.inv(self.poses[seq][i_img]), self.poses[seq][i_img + 1])
m_forward_mirror = np.dot(np.linalg.inv(mirror_pose), mirror_pose_next)
m_reverse = np.dot(np.linalg.inv(self.poses[seq][i_img + 1]), self.poses[seq][i_img])
m_reverse_mirror = np.dot(np.linalg.inv(mirror_pose_next), mirror_pose)
trans_forward = transformations.translation_from_matrix(m_forward)
ypr_forward = transformations.euler_from_matrix(m_forward, axes="rzyx")
trans_forward_mirror = transformations.translation_from_matrix(m_forward_mirror)
ypr_forward_mirror = transformations.euler_from_matrix(m_forward_mirror, axes="rzyx")
trans_reverse = transformations.translation_from_matrix(m_reverse)
ypr_reverse = transformations.euler_from_matrix(m_reverse, axes="rzyx")
trans_reverse_mirror = transformations.translation_from_matrix(m_reverse_mirror)
ypr_reverse_mirror = transformations.euler_from_matrix(m_reverse_mirror, axes="rzyx")
self.fc_ground_truth[seq][i_img] = np.concatenate([trans_forward, ypr_forward])
self.fc_mirror_ground_truth[seq][i_img] = np.concatenate([trans_forward_mirror, ypr_forward_mirror])
self.fc_reverse_ground_truth[seq][i_img] = np.concatenate([trans_reverse, ypr_reverse])
self.fc_reverse_mirror_ground_truth[seq][i_img] = np.concatenate(
[trans_reverse_mirror, ypr_reverse_mirror])
get_imu = seq_loader.get_averaged_imu_in_corresponding_frame
self.imu_measurements[seq][i_img] = get_imu(i_img, mirror=False, reverse=False)
self.imu_measurements_mirror[seq][i_img] = get_imu(i_img, mirror=True, reverse=False)
self.imu_measurements_reverse[seq][i_img] = get_imu(i_img, mirror=False, reverse=True)
self.imu_measurements_reverse_mirror[seq][i_img] = get_imu(i_img, mirror=True, reverse=True)
# How many examples the sequence contains, were it to be processed using a batch size of 1
sequence_examples = np.ceil((num_frames - self.cfg.timesteps) / self.cfg.sequence_stride).astype(
np.int32)
# how many batches in the sequence, cutting off any extra
if self.cfg.bidir_aug:
self.batch_counts[seq] = np.floor(2 * sequence_examples / self.cfg.batch_size).astype(np.int32)
else:
self.batch_counts[seq] = np.floor(sequence_examples / self.cfg.batch_size).astype(np.int32)
self.batch_cnt += self.batch_counts[seq].astype(np.int32)
self.batch_order[seq] = np.arange(0, self.batch_counts[seq], dtype=np.uint32)
self.batch_offsets[seq] = np.zeros([self.cfg.batch_size], dtype=np.uint32)
self.sequence_batch[seq] = 0
self.total_frames += num_frames
tools.printf("All data loaded, batch_size=%d, timesteps=%d, num_batches=%d" % (
self.cfg.batch_size, self.cfg.timesteps, self.batch_cnt))
gc.collect()
self.sequence_ordering = np.zeros([self.batch_cnt], dtype=np.uint16)
self.set_batch_offsets()
self.next_epoch(False)
def translation_from_matrix(m):
return tf.translation_from_matrix(m)
def runVIO(veh, callback):
global vehicle, body_offset_enabled, body_offset_x, body_offset_y, body_offset_z
global compass_enabled, scale_calib_enable, scale_factor, camera_orientation, pipe, is_vehicle_connected
global home_lat, home_lon, home_alt, auto_set_ekf_home_enable, debug_enable
global H_aeroRef_T265Ref, H_T265body_aeroBody, data, current_confidence, H_aeroRef_aeroBody, heading_north_yaw
global current_time
print("runVIO() started")
vehicle = veh
is_vehicle_connected = True
if body_offset_enabled == 1:
print("INFO: Using camera position offset: Enabled, x y z is",
body_offset_x, body_offset_y, body_offset_z)
else:
print("INFO: Using camera position offset: Disabled")
if compass_enabled == 1:
print(
"INFO: Using compass: Enabled. Heading will be aligned to north.")
else:
print("INFO: Using compass: Disabled")
if scale_calib_enable == True:
print(
"\nINFO: SCALE CALIBRATION PROCESS. DO NOT RUN DURING FLIGHT.\nINFO: TYPE IN NEW SCALE IN FLOATING POINT FORMAT\n"
)
else:
if scale_factor == 1.0:
print("INFO: Using default scale factor", scale_factor)
else:
print("INFO: Using scale factor", scale_factor)
print("INFO: Using camera orientation", camera_orientation)
# Transformation to convert different camera orientations to NED convention. Replace camera_orientation_default for your configuration.
# 0: Forward, USB port to the right
# 1: Downfacing, USB port to the right
# 2: Forward, 45 degree tilted down
# Important note for downfacing camera: you need to tilt the vehicle's nose up a little - not flat - before you run the script, otherwise the initial yaw will be randomized, read here for more details: https://github.com/IntelRealSense/librealsense/issues/4080. Tilt the vehicle to any other sides and the yaw might not be as stable.
if camera_orientation == 0:
# Forward, USB port to the right
H_aeroRef_T265Ref = np.array([[0, 0, -1, 0], [1, 0, 0, 0],
[0, -1, 0, 0], [0, 0, 0, 1]])
H_T265body_aeroBody = np.linalg.inv(H_aeroRef_T265Ref)
elif camera_orientation == 1:
# Downfacing, USB port to the right
H_aeroRef_T265Ref = np.array([[0, 0, -1, 0], [1, 0, 0, 0],
[0, -1, 0, 0], [0, 0, 0, 1]])
H_T265body_aeroBody = np.array([[0, 1, 0, 0], [1, 0, 0, 0],
[0, 0, -1, 0], [0, 0, 0, 1]])
elif camera_orientation == 2:
# 45degree forward
H_aeroRef_T265Ref = np.array([[0, 0, -1, 0], [1, 0, 0, 0],
[0, -1, 0, 0], [0, 0, 0, 1]])
H_T265body_aeroBody = np.array([[0., 1., 0., 0.],
[-0.70710676, -0., -0.70710676, -0.],
[-0.70710676, 0., 0.70710676, 0.],
[0., 0., 0., 1.]])
else:
# Default is facing forward, USB port to the right
H_aeroRef_T265Ref = np.array([[0, 0, -1, 0], [1, 0, 0, 0],
[0, -1, 0, 0], [0, 0, 0, 1]])
H_T265body_aeroBody = np.linalg.inv(H_aeroRef_T265Ref)
if auto_set_ekf_home_enable == False:
print("INFO: Automatically set EKF home: DISABLED")
else:
print("INFO: Automatically set EKF home: Enabled")
if not debug_enable:
debug_enable = 0
else:
debug_enable = 1
np.set_printoptions(precision=4,
suppress=True) # Format output on terminal
print("INFO: Debug messages enabled.")
#######################################
# Main code starts here
#######################################
print("INFO: Connecting to Realsense camera.")
realsense_connect()
print("INFO: Realsense connected.")
# Listen to the mavlink messages that will be used as trigger to set EKF home automatically
if auto_set_ekf_home_enable == True:
vehicle.add_message_listener('STATUSTEXT', statustext_callback)
if compass_enabled == 1:
# Listen to the attitude data in aeronautical frame
vehicle.add_message_listener('ATTITUDE', att_msg_callback)
data = None
current_confidence = None
H_aeroRef_aeroBody = None
heading_north_yaw = None
# Send MAVlink messages in the background
sched = BackgroundScheduler()
sched.add_job(send_vision_position_message,
'interval',
seconds=1 / vision_msg_hz)
sched.add_job(send_confidence_level_dummy_message,
'interval',
seconds=1 / confidence_msg_hz)
# For scale calibration, we will use a thread to monitor user input
if scale_calib_enable == True:
scale_update_thread = threading.Thread(target=scale_update)
scale_update_thread.daemon = True
scale_update_thread.start()
sched.start()
if compass_enabled == 1:
# Wait a short while for yaw to be correctly initiated
time.sleep(1)
print("INFO: Sending VISION_POSITION_ESTIMATE messages to FCU.")
callback()
try:
while True:
# Wait for the next set of frames from the camera
frames = pipe.wait_for_frames()
# Fetch pose frame
pose = frames.get_pose_frame()
if pose:
# Store the timestamp for MAVLink messages
current_time = int(round(time.time() * 1000000))
# Pose data consists of translation and rotation
data = pose.get_pose_data()
# In transformations, Quaternions w+ix+jy+kz are represented as [w, x, y, z]!
H_T265Ref_T265body = tf.quaternion_matrix([
data.rotation.w, data.rotation.x, data.rotation.y,
data.rotation.z
])
H_T265Ref_T265body[0][3] = data.translation.x * scale_factor
H_T265Ref_T265body[1][3] = data.translation.y * scale_factor
H_T265Ref_T265body[2][3] = data.translation.z * scale_factor
# Transform to aeronautic coordinates (body AND reference frame!)
H_aeroRef_aeroBody = H_aeroRef_T265Ref.dot(
H_T265Ref_T265body.dot(H_T265body_aeroBody))
# Take offsets from body's center of gravity (or IMU) to camera's origin into account
if body_offset_enabled == 1:
H_body_camera = tf.euler_matrix(0, 0, 0, 'sxyz')
H_body_camera[0][3] = body_offset_x
H_body_camera[1][3] = body_offset_y
H_body_camera[2][3] = body_offset_z
H_camera_body = np.linalg.inv(H_body_camera)
H_aeroRef_aeroBody = H_body_camera.dot(
H_aeroRef_aeroBody.dot(H_camera_body))
# Realign heading to face north using initial compass data
if compass_enabled == 1:
H_aeroRef_aeroBody = H_aeroRef_aeroBody.dot(
tf.euler_matrix(0, 0, heading_north_yaw, 'sxyz'))
# Show debug messages here
if debug_enable == 1:
os.system(
'clear'
) # This helps in displaying the messages to be more readable
print("DEBUG: Raw RPY[deg]: {}".format(
np.array(
tf.euler_from_matrix(H_T265Ref_T265body, 'sxyz')) *
180 / m.pi))
print("DEBUG: NED RPY[deg]: {}".format(
np.array(
tf.euler_from_matrix(H_aeroRef_aeroBody, 'sxyz')) *
180 / m.pi))
print("DEBUG: Raw pos xyz : {}".format(
np.array([
data.translation.x, data.translation.y,
data.translation.z
])))
print("DEBUG: NED pos xyz : {}".format(
np.array(
tf.translation_from_matrix(H_aeroRef_aeroBody))))
except KeyboardInterrupt:
print("T265: INFO: KeyboardInterrupt has been caught. Cleaning up...")
finally:
pipe.stop()
print("INFO: Realsense pipeline and vehicle object closed.")
sys.exit()
def position(self): ''' @brief Extract branch origin position from the transformation matrix @return a 3D position vector ''' return tr.translation_from_matrix(self.frame)
def marker_hover(vehicle, marker_tracker, rs=None):
pid_x = PID(1, 0, 0, setpoint=0)
pid_y = PID(1, 0, 0, setpoint=0)
pid_z = PID(1, 0, 0, setpoint=-0.5)
pid_x.sample_time = 0.01
pid_y.sample_time = 0.01
pid_z.sample_time = 0.01
# Hover until manual override
pose_queue = np.zeros((RUNNING_AVG_LENGTH, 3), dtype=float)
count = 0
if args["debug"] > 0:
vehicle_pose_file = file_utils.open_file(VEHICLE_POSE_FILE)
start_time = time.time()
print("Tracking marker...")
while True:
# while vehicle.mode == VehicleMode("GUIDED"):
# Track marker
marker_tracker.track_marker(
alt=vehicle.location.global_relative_frame.alt)
if marker_tracker.is_marker_found():
marker_pose = marker_tracker.get_pose()
# Transform pose to UAV body frame; proportional gain
marker_pose_body_ref = marker_ref_to_body_ref(marker_pose)
# Pid response, input is UAV rel. to marker
control_pose = np.array([[
pid_x(-marker_pose_body_ref[0]),
pid_y(-marker_pose_body_ref[1]),
pid_z(-marker_pose_body_ref[2])
]])
# Keep a queue of recent marker poses
pose_queue[count % len(pose_queue)] = control_pose
count += 1
if count < RUNNING_AVG_LENGTH - 1:
pose_avg = np.sum(pose_queue, axis=0) / count
else:
pose_avg = np.average(pose_queue, axis=0)
# Send position command to the vehicle
dronekit_utils.goto_position_target_body_offset_ned(
vehicle,
forward=pose_avg[0],
right=pose_avg[1],
down=-HOVER_ALTITUDE - pose_avg[2])
if args["debug"] > 0 and rs is not None:
data = rs.read()
rs_pose = tf.quaternion_matrix([
data.rotation.w, data.rotation.x, data.rotation.y,
data.rotation.z
])
rs_pose[0][3] = data.translation.x
rs_pose[1][3] = data.translation.y
rs_pose[2][3] = data.translation.z
vehicle_pose = realsense_localization.rs_to_body(rs_pose)
vehicle_trans = np.array(
tf.translation_from_matrix(vehicle_pose))
vehicle_pose_file.write("{} {} {} {} 0 0 0 0\n".format(
time.time() - start_time, vehicle_trans[0],
vehicle_trans[1], vehicle_trans[2]))
if args["debug"] > 2:
print("Nav command: {} {} {}".format(pose_avg[0], pose_avg[1],
pose_avg[2]))
# Maintain frequency of sending commands
marker_tracker.wait()
def saveiqe(model, filename):
model['bone_translate'] = []
model['bone_rotation'] = []
bone_count = len(model['bone_parent'])
for i in range(bone_count):
matrix = tf.identity_matrix()
parent_node = model['bone_parent'][i]
if parent_node >= 0:
matrix = model['bone_original_matrix'][parent_node]
matrix = np.dot(model['bone_original_matrix'][i],
np.linalg.inv(matrix))
model['bone_translate'].append(tf.translation_from_matrix(matrix.T))
model['bone_rotation'].append(tf.quaternion_from_matrix(matrix.T))
with open(filename + '.iqe ', 'w') as f:
f.write('# Inter-Quake Export\n')
f.write('\n')
parent_child_dict = {}
old2new = {}
index_pool = [-1]
for i in range(len(model['bone_parent'])):
p = model['bone_parent'][i]
if p not in parent_child_dict:
parent_child_dict[p] = []
parent_child_dict[p].append(i)
def print_joint(index, parent_index):
f.write('joint "{}" {}\n'.format(model['bone_name'][index],
parent_index))
x, y, z = model['bone_translate'][index]
r, i, j, k = model['bone_rotation'][index]
f.write('pq {} {} {} {} {} {} {}\n'.format(-x, y, z, i, -j, -k, r))
def deep_first_search(index, index_pool, parent_index):
index_pool[0] += 1
current_node_index = index_pool[0]
old2new[index] = current_node_index
print_joint(index, parent_index)
if index in parent_child_dict:
for child in parent_child_dict[index]:
deep_first_search(child, index_pool, current_node_index)
deep_first_search(model['bone_parent'].index(-1), index_pool, -1)
f.write('\n')
mesh_vertex_counter = 0
mesh_face_counter = 0
for mesh_i in range(len(model['mesh'])):
mesh_vertex_counter_end = mesh_vertex_counter + model['mesh'][
mesh_i][0]
mesh_face_counter_end = mesh_face_counter + model['mesh'][mesh_i][1]
f.write('mesh mesh{}\n'.format(mesh_i))
f.write('material "mesh{}Mat"\n'.format(mesh_i))
f.write('\n')
for i in range(mesh_vertex_counter, mesh_vertex_counter_end):
x, y, z = model['position'][i]
f.write('vp {} {} {}\n'.format(-x, y, z))
f.write('\n')
for i in range(mesh_vertex_counter, mesh_vertex_counter_end):
x, y, z = model['normal'][i]
f.write('vn {} {} {}\n'.format(-x, y, z))
f.write('\n')
for i in range(mesh_vertex_counter, mesh_vertex_counter_end):
u, v = model['uv'][i]
f.write('vt {} {}\n'.format(u, 1 - v))
f.write('\n')
for i in range(mesh_vertex_counter, mesh_vertex_counter_end):
f.write('vb')
for j in range(4):
v = model['vertex_joint'][i][j]
if v == 255:
break
v = old2new[v]
w = model['vertex_joint_weight'][i][j]
f.write(' {} {}'.format(v, w))
f.write('\n')
f.write('\n')
for i in range(mesh_face_counter, mesh_face_counter_end):
v1, v2, v3 = model['face'][i]
v1 -= mesh_vertex_counter
v2 -= mesh_vertex_counter
v3 -= mesh_vertex_counter
f.write('fm {} {} {}\n'.format(v3, v1, v2))
f.write('\n')
mesh_vertex_counter = mesh_vertex_counter_end
mesh_face_counter = mesh_face_counter_end
def __init__(self, config, base_dir, sequences, frames=None):
self.truncated_seq_sizes = []
self.end_of_sequence_indices = []
self.curr_batch_idx = 0
self.unused_batch_indices = []
self.cfg = config
if not frames:
frames = [None] * len(sequences)
total_num_examples = 0
for i_seq, seq in enumerate(sequences):
seq_data = pykitti.odometry(base_dir, seq, frames=frames[i_seq])
num_frames = len(seq_data.poses)
# less than timesteps number of frames will be discarded
num_examples = (num_frames - 1) // self.cfg.timesteps
self.truncated_seq_sizes.append(num_examples * self.cfg.timesteps + 1)
total_num_examples += num_examples
# less than batch size number of examples will be discarded
self.total_batch_count = total_num_examples // self.cfg.batch_size
# +1 adjusts for the extra image in the last time step
total_timesteps = self.total_batch_count * (self.cfg.timesteps + 1)
# since some examples will be discarded, readjust the truncated_seq_sizes
deleted_frames = (total_num_examples - self.total_batch_count * self.cfg.batch_size) * self.cfg.timesteps
for i in range(len(self.truncated_seq_sizes) - 1, -1, -1):
if self.truncated_seq_sizes[i] > deleted_frames:
self.truncated_seq_sizes[i] -= deleted_frames
break
else:
self.truncated_seq_sizes[i] = 0
deleted_frames -= self.truncated_seq_sizes[i]
# for storing all training
self.input_frames = np.zeros(
[total_timesteps, self.cfg.batch_size, self.cfg.input_channels, self.cfg.input_height,
self.cfg.input_width],
dtype=np.uint8)
poses_wrt_g = np.zeros([total_timesteps, self.cfg.batch_size, 4, 4], dtype=np.float32) # ground truth poses
num_image_loaded = 0
for i_seq, seq in enumerate(sequences):
seq_data = pykitti.odometry(base_dir, seq, frames=frames[i_seq])
length = self.truncated_seq_sizes[i_seq]
i = -1
j = -1
for i_img in range(length):
if i_img % 100 == 0:
tools.printf("Loading sequence %s %.1f%% " % (seq, (i_img / length) * 100))
i = num_image_loaded % total_timesteps
j = num_image_loaded // total_timesteps
# swap axis to channels first
img = seq_data.get_cam2(i_img)
img = img.resize((self.cfg.input_width, self.cfg.input_height))
img = np.array(img)
if img.shape[2] == 3: # convert to bgr if colored image
img = img[..., [2, 1, 0]]
img = np.reshape(img, [img.shape[0], img.shape[1], self.cfg.input_channels])
img = np.moveaxis(np.array(img), 2, 0)
pose = seq_data.poses[i_img]
self.input_frames[i, j] = img
poses_wrt_g[i, j] = pose
num_image_loaded += 1
# insert the extra image at the end of the batch, note the number of
# frames per batch of batch size 1 is timesteps + 1
if i_img != 0 and i_img != length - 1 and i_img % self.cfg.timesteps == 0:
i = num_image_loaded % total_timesteps
j = num_image_loaded // total_timesteps
self.input_frames[i, j] = img
poses_wrt_g[i, j] = pose
num_image_loaded += 1
# If the batch has the last frame in a sequence, the following frame
# in the next batch must have a reset for lstm state
self.end_of_sequence_indices.append((i + 1, j,))
# make sure the all of examples are fully loaded, just to detect bugs
assert (num_image_loaded == total_timesteps * self.cfg.batch_size)
# now convert all the ground truth from 4x4 to xyz + quat, this is after the SE3 layer
self.se3_ground_truth = np.zeros([total_timesteps, self.cfg.batch_size, 7], dtype=np.float32)
for i in range(0, self.se3_ground_truth.shape[0]):
for j in range(0, self.se3_ground_truth.shape[1]):
translation = transformations.translation_from_matrix(poses_wrt_g[i, j])
quat = transformations.quaternion_from_matrix(poses_wrt_g[i, j])
self.se3_ground_truth[i, j] = np.concatenate([translation, quat])
# extract the relative transformation between frames after the fully connected layer
self.fc_ground_truth = np.zeros([total_timesteps, self.cfg.batch_size, 6], dtype=np.float32)
# going through rows, then columns
for i in range(0, self.fc_ground_truth.shape[0]):
for j in range(0, self.fc_ground_truth.shape[1]):
# always identity at the beginning of the sequence
if i % (self.cfg.timesteps + 1) == 0:
m = transformations.identity_matrix()
else:
m = np.dot(np.linalg.inv(poses_wrt_g[i - 1, j]), poses_wrt_g[i, j]) # double check
translation = transformations.translation_from_matrix(m)
ypr = transformations.euler_from_matrix(m, axes="rzyx")
assert (np.all(np.abs(ypr) < np.pi))
self.fc_ground_truth[i, j] = np.concatenate([translation, ypr]) # double check
tools.printf("All data loaded, batch_size=%d, timesteps=%d, num_batches=%d" % (
self.cfg.batch_size, self.cfg.timesteps, self.total_batch_count))
self.next_epoch()
gc.collect() # force garbage collection
def localize(rs, sched=None):
# Global variables are usually not good style, but are required here (for now)
global vehicle, H0_2, data, current_time
# Listen to the mavlink messages that will be used as trigger to set the EKF home automatically
# The Pixhawk will not allow arming if the EKF and Home origins have not been set. Once data is
# being received from the Realsense, this callback will set the origins.
# NOTE: There has been some trouble with this; sometimes it fails to set the origin. If this happens,
# reboot the Pixhawk until the issue resolves itself (or find a better solution!)
vehicle.add_message_listener('STATUSTEXT', statustext_callback)
if compass_enabled == 1:
# Listen to the attitude data in aeronautical frame
vehicle.add_message_listener('ATTITUDE', att_msg_callback)
# Send MAVlink messages in the background
if sched is None:
sched = BackgroundScheduler()
# Add message senders for vision pose message and confidence message
sched.add_job(send_vision_position_message,
'interval',
seconds=1 / vision_msg_hz,
max_instances=5)
sched.add_job(send_confidence_level_dummy_message,
'interval',
seconds=1 / confidence_msg_hz)
# For scale calibration, we will use a thread to monitor user input
# NOTE: not tested by VADL
if scale_calib_enable:
scale_update_thread = threading.Thread(target=scale_update)
scale_update_thread.daemon = True
scale_update_thread.start()
# Start sending background messages to Mavlink
sched.start()
if compass_enabled == 1:
# Wait a short while for yaw to be correctly initiated
time.sleep(1)
# This function is contained in its own thread and runs indefinitely.
start_time = time.time()
print("Sending VISION_POSITION_ESTIMATE messages to FCU.")
try:
while True:
# Obtain pose from the Realsense
data = rs.read()
if data:
# Store the timestamp for MAVLink messages
current_time = int(round(time.time() * 1000000))
# Extract pose info and format as a homogeneous matrix.
# In transformations, Quaternions w+ix+jy+kz are represented as [w, x, y, z]
H1_3 = tf.quaternion_matrix([
data.rotation.w, data.rotation.x, data.rotation.y,
data.rotation.z
])
H1_3[0][3] = data.translation.x * scale_factor
H1_3[1][3] = data.translation.y * scale_factor
H1_3[2][3] = data.translation.z * scale_factor
# Transform to NED coordinates (body AND reference frame!)
H0_2 = rs_to_body(H1_3)
# Account for offsets from body's center of gravity (or IMU) to camera's origin
if body_offset_enabled == 1:
H_body_camera = tf.euler_matrix(0, 0, 0, 'sxyz')
H_body_camera[0][3] = body_offset_x
H_body_camera[1][3] = body_offset_y
H_body_camera[2][3] = body_offset_z
H_camera_body = np.linalg.inv(H_body_camera)
# Shift the NED pose by an offset
H0_2 = H_body_camera.dot(H0_2.dot(H_camera_body))
# Record time
timestamp = time.time() - start_time
# Convert to NED & quaternion for data logging
pose_ned = np.array(tf.translation_from_matrix(H0_2))
# Write pose and accelerations to file
# These logs follow the .tum format so they can be processed by
# the EVO SLAM comparisons package:
#
# https://github.com/MichaelGrupp/evo
rs_pose_file.write(
str(timestamp) + " " + str(pose_ned[0]) + " " +
str(pose_ned[1]) + " " + str(pose_ned[2]) + " " +
str(data.rotation.w) + " " + str(data.rotation.x) + " " +
str(data.rotation.y) + " " + str(data.rotation.z) + "\n")
rs_accel_file.write(
str(timestamp) + " " + str(data.acceleration.x) + " " +
str(data.acceleration.y) + " " + str(data.acceleration.z) +
"\n")
# Show debug messages here
if debug_enable == 1:
os.system(
'clear'
) # This helps in displaying the messages to be more readable
print("DEBUG: Raw RPY[deg]: {}".format(
np.array(tf.euler_from_matrix(H1_3, 'sxyz')) * 180 /
m.pi))
print("DEBUG: NED RPY[deg]: {}".format(
np.array(tf.euler_from_matrix(H0_2, 'sxyz')) * 180 /
m.pi))
print("DEBUG: Raw pos xyz : {}".format(
np.array([
data.translation.x, data.translation.y,
data.translation.z
])))
print("DEBUG: NED pos xyz : {}".format(
np.array(tf.translation_from_matrix(H0_2))))
except KeyboardInterrupt:
print("EXCEPTION: KeyboardInterrupt has been caught. Cleaning up...")
finally:
# Clean up Realsense and Vehicle objects
rs.stop()
vehicle.close()
print("Realsense pipeline and vehicle object closed.")
sys.exit()
def _pack(T):
eulxyz = tf.euler_from_matrix(T, axes='sxyz')
trans = tf.translation_from_matrix(T)
return np.hstack((trans, eulxyz))
progress("DEBUG: Raw RPY[deg]: {}".format(
np.array(
tf.euler_from_matrix(H_T265Ref_T265body, 'sxyz')) *
180 / m.pi))
progress("DEBUG: NED RPY[deg]: {}".format(
np.array(
tf.euler_from_matrix(H_aeroRef_aeroBody, 'sxyz')) *
180 / m.pi))
progress("DEBUG: Raw pos xyz : {}".format(
np.array([
data.translation.x, data.translation.y,
data.translation.z
])))
progress("DEBUG: NED pos xyz : {}".format(
np.array(
tf.translation_from_matrix(H_aeroRef_aeroBody))))
conn.home_location = LocationGlobal(home_lat, home_lon, home_alt)
arm_and_takeoff(0)
except Exception as e:
progress(e)
except:
send_msg_to_gcs('ERROR IN SCRIPT')
progress("Unexpected error: %s" % sys.exc_info()[0])
finally:
progress('Closing the script...')
# start a timer in case stopping everything nicely doesn't work.
signal.setitimer(signal.ITIMER_REAL, 5) # seconds...
def tf_to_pose_stamp(tf, id):
q = transformations.quaternion_from_matrix(tf)
t = transformations.translation_from_matrix(tf)
return create_pose_stamp_msg(id, [t[0], t[1], t[2], q[0], q[1], q[2], q[3]])