def getEnv(self, key, iter=0, first=False):
val = self.base.getModel(key)
print key, val
obj = None
obj_list = []
parent_tf = tf.identity_matrix()
if self.base.typeOf(key) == self.base.ROBOTS.ident:
obj = self._getRobot(key)
obj_list.append(obj)
parent_tf = obj.GetTransform()
elif self.base.typeOf(key) == self.base.OBJECTS.ident:
obj = self._getObject(key)
obj_list.append(obj)
parent_tf = obj.GetTransform()
elif self.base.typeOf(key) == self.base.LOCATIONS.ident:
obj = self._getLocation(key)
obj_list.append(obj)
parent_tf = obj.GetTransform()
elif self.base.typeOf(key) == self.base.SENSORS.ident:
obj = self._getSensor(key)
obj_list.append(obj)
parent_tf = obj.GetTransform()
else:
if val.has_key("translation"):
translation = map(lambda x: float(x), val["translation"].split(" "))
parent_tf = tf.concatenate_matrices(parent_tf, tf.compose_matrix(translate = translation))
if val.has_key("quat"):
quat = map(lambda x: float(x), val["quat"].split(" "))
rot = rave.axisAngleFromQuat(quat)
m2 = tf.compose_matrix(angles = rot)
parent_tf = tf.concatenate_matrices(parent_tf, m2)
if first:
parent_tf = tf.identity_matrix()
if obj != None:
obj.SetTransform(parent_tf)
if iter==0:
return obj_list
# search for ancestors
child_expr = pycassa.create_index_expression('base', key)
clild_clause = pycassa.create_index_clause([child_expr])
for child_key, _ in self.base.col.get_indexed_slices(clild_clause):
child_obj = self.getEnv(child_key, iter-1)
for obj in child_obj:
if type(obj) != type(None):
obj.SetTransform(tf.concatenate_matrices(parent_tf, obj.GetTransform()))
obj_list.append(obj)
return obj_list
def motion_model(self, f0, f1, stamp, use_kalman=False):
if not use_kalman:
# simple `repetition` model
txn0, rxn0 = f0['pose'][L_POS], f0['pose'][A_POS]
txn1, rxn1 = f1['pose'][L_POS], f1['pose'][A_POS]
R0 = tx.euler_matrix(*rxn0)
R1 = tx.euler_matrix(*rxn1)
T0 = tx.compose_matrix(angles=rxn0, translate=txn0)
T1 = tx.compose_matrix(angles=rxn1, translate=txn1)
Tv = np.dot(T1, vm.inv(T0)) # Tv * T0 = T1
T2 = np.dot(Tv, T1)
txn = tx.translation_from_matrix(T2)
rxn = tx.euler_from_matrix(T2)
x = f1['pose'].copy()
P = f1['cov'].copy()
x[0:3] = txn
x[9:12] = rxn
return x, P
else:
# dt MUST NOT BE None
self.kf_.x = f0['pose']
self.kf_.P = f0['cov']
dt = (f1['stamp'] - f0['stamp'])
self.kf_.predict(dt)
txn, rxn = f1['pose'][L_POS], f1['pose'][A_POS]
z = np.concatenate([txn, rxn])
self.kf_.update(z)
dt = (stamp - f1['stamp'])
self.kf_.predict(dt)
return self.kf_.x.copy(), self.kf_.P.copy()
def pose_callback(self, msg):
print "pose cb"
trans = [msg.pose.position.x, msg.pose.position.y, msg.pose.position.z]
rot = [
msg.pose.orientation.x, msg.pose.orientation.y,
msg.pose.orientation.z, msg.pose.orientation.w
]
pose_matrix = tr.compose_matrix(angles=tr.euler_from_quaternion(rot),
translate=trans)
# Transfer to the frame of the robot
# pose_matrix = np.dot(self.config, pose_matrix)
if self.previous_tf is None: #if first callback
odo_trans = trans
odo_rot = rot
odo = tr.compose_matrix(angles=tr.euler_from_quaternion(odo_rot),
translate=odo_trans)
self.origin = trans
else:
# calculate limit matrix
if enable_filter:
limit_dist = limit_speed * (msg.header.stamp.to_nsec() -
self.previous_time.to_nsec())
print(msg.header.stamp.to_nsec() -
self.previous_time.to_nsec())
scale, shear, angles, prev_trans, persp = tr.decompose_matrix(
self.previous_tf)
moved_vec = [
msg.pose.position.x - prev_trans[0],
msg.pose.position.y - prev_trans[1],
msg.pose.position.z - prev_trans[2]
]
moved_dist = np.linalg.norm(moved_vec)
if moved_dist > limit_dist:
#discard this pose
print "move too much"
return
odo = np.dot(tr.inverse_matrix(self.previous_tf), pose_matrix)
odo_trans = tf.transformations.translation_from_matrix(odo)
odo_rot = tf.transformations.quaternion_from_matrix(odo)
self.previous_time = msg.header.stamp
self.previous_tf = pose_matrix
print "x: ", trans[0] - self.origin[0], "y: ", trans[1] - self.origin[
1], "z: ", trans[2] - self.origin[2]
robot_odo = PoseStamped()
robot_odo.header.stamp = msg.header.stamp
robot_odo.pose.position.x = odo_trans[0]
robot_odo.pose.position.y = odo_trans[1]
robot_odo.pose.position.z = odo_trans[2]
robot_odo.pose.orientation.x = odo_rot[0]
robot_odo.pose.orientation.y = odo_rot[1]
robot_odo.pose.orientation.z = odo_rot[2]
robot_odo.pose.orientation.w = odo_rot[3]
self.pub_odom.publish(robot_odo)
def updateStoredTransforms(update):
global holoTransformList
global markerTransformList
global worldTransform
global offsetTransform
arrSize = holoTransformList.shape[0]
#parse input into meaningful values
xworldT = update[0:3]
xworldR = update[3:6]
xworldRQ = TR.quaternion_about_axis(np.sqrt(xworldR.dot(xworldR)), xworldR)
xoffsetT = update[6:9]
xoffsetR = update[9:12]
xoffsetRQ = TR.quaternion_about_axis(np.sqrt(xoffsetR.dot(xoffsetR)),
xoffsetR)
#update world transform
worldIncrement = TR.compose_matrix(
translate=xworldT, angles=TR.euler_from_quaternion(xworldRQ))
worldTransform = np.dot(worldIncrement, worldTransform)
#update offset transform
offsetIncrement = TR.compose_matrix(
translate=xoffsetT, angles=TR.euler_from_quaternion(xoffsetRQ))
offsetTransform = np.dot(offsetIncrement, offsetTransform)
def point_cb(self, msg):
print "point cb"
trans = [
self.pose_msg.pose.position.x / 1000,
self.pose_msg.pose.position.y / 1000,
self.pose_msg.pose.position.z / 1000
]
rot = [
self.pose_msg.pose.orientation.x, self.pose_msg.pose.orientation.y,
self.pose_msg.pose.orientation.z, self.pose_msg.pose.orientation.w
]
pose_matrix = tr.compose_matrix(angles=tr.euler_from_quaternion(rot),
translate=trans)
pt_trans = [msg.point.x, msg.point.y, msg.point.z]
pt_rot = [0, 0, 0, 1]
pt_matrix = tr.compose_matrix(angles=tr.euler_from_quaternion(pt_rot),
translate=pt_trans)
obj_pose = np.dot(pose_matrix, pt_matrix)
obj_position = tf.transformations.translation_from_matrix(obj_pose)
print "pose = ", self.pose_msg.pose
print "point = ", msg.point
print "obj position = ", obj_position
point_msg = PointStamped()
point_msg.header.stamp = rospy.Time()
point_msg.header.seq = msg.header.seq
point_msg.header.frame_id = "map"
point_msg.point.x = obj_position[0]
point_msg.point.y = obj_position[1]
point_msg.point.z = obj_position[2]
self.pub_obj_position.publish(point_msg)
def addTransforms(self, frame1, frame2):
""" Adds two transform objects to get a resulting transform. """
t1 = np.array(
[frame1.translation.x, frame1.translation.y, frame1.translation.z])
t2 = np.array(
[frame2.translation.x, frame2.translation.y, frame2.translation.z])
r1 = np.array([
frame1.rotation.x, frame1.rotation.y, frame1.rotation.z,
frame1.rotation.w
])
r2 = np.array([
frame2.rotation.x, frame2.rotation.y, frame2.rotation.z,
frame2.rotation.w
])
f1 = t.compose_matrix(translate=t1, angles=t.euler_from_quaternion(r1))
f2 = t.compose_matrix(translate=t2, angles=t.euler_from_quaternion(r2))
f = np.matmul(f1, f2)
trans = t.translation_from_matrix(f)
rot = t.quaternion_from_matrix(f)
trans = Vector3(trans[0], trans[1], trans[2])
rot = Quaternion(rot[0], rot[1], rot[2], rot[3])
return Transform(trans, rot)
def angle_diff_from_quaternions(q1, q2):
tf1 = transformations.compose_matrix(
angles=transformations.euler_from_quaternion(q1))
tf2 = transformations.compose_matrix(
angles=transformations.euler_from_quaternion(q2))
angle, _, _ = transformations.rotation_from_matrix(
np.dot(np.linalg.inv(tf1), tf2))
return angle / np.pi * 180.
def _get_tf_matrix(self, pose):
if isinstance(pose, dict):
_quat = np.array([pose['orientation'].x, pose['orientation'].y, pose['orientation'].z, pose['orientation'].w])
_trans = np.array([pose['position'].x, pose['position'].y, pose['position'].z])
return tr.compose_matrix(angles=tr.euler_from_quaternion(_quat,'sxyz'), translate=_trans)
else:
_quat = np.array([pose.orientation.x, pose.orientation.y, pose.orientation.z, pose.orientation.w])
_trans = np.array([pose.position.x, pose.position.y, pose.position.z])
return tr.compose_matrix(angles=tr.euler_from_quaternion(_quat,'sxyz'), translate=_trans)
def transform_pose_to_base_link(self, t, q):
euler_pose = tfm.euler_from_quaternion(q)
tf_cam_col_opt_fram = tfm.compose_matrix(translate=t,
angles=euler_pose)
trans, quat = self.listener.lookupTransform(
'base_link', 'camera_color_optical_frame', rospy.Time(0))
euler = tfm.euler_from_quaternion(quat)
tf = tfm.compose_matrix(translate=trans, angles=euler)
t_pose = np.dot(tf, tf_cam_col_opt_fram)
t_ba_li = t_pose[0:3, 3]
return t_ba_li
def transformOpenrave(self, openrave_obj, complex_data):
if complex_data.has_key("translation"):
translation = map(lambda x: float(x), complex_data["translation"].split(" "))
m1 = openrave_obj.GetTransform()
m2 = tf.compose_matrix(translate = translation)
openrave_obj.SetTransform(tf.concatenate_matrices(m1, m2))
if complex_data.has_key("quat"):
quat = map(lambda x: float(x), complex_data["quat"].split(" "))
m1 = openrave_obj.GetTransform()
rot = rave.axisAngleFromQuat(quat)
m2 = tf.compose_matrix(angles = rot)
openrave_obj.SetTransform(tf.concatenate_matrices(m1, m2))
return openrave_obj
def controller():
rospy.sleep(1)
print("en el controler")
k_v = 0.4
k_w = 1
x_goal = -3
y_goal = 2
theta_goal = 0
while not rospy.is_shutdown():
print("true goal:", x_goal, y_goal)
(roll_bot, pitch_bot, yaw_bot) = euler_from_quaternion([
pose.orientation.x, pose.orientation.y, pose.orientation.z,
pose.orientation.w
])
orient = np.array([roll_bot, pitch_bot, yaw_bot])
trans = np.array([pose.position.x, pose.position.y, 0])
T_bot = compose_matrix(angles=orient, translate=trans)
T_bot_inv = np.linalg.inv(T_bot)
orient = np.array([0, 0, 0])
trans = np.array([x_goal, y_goal, 0])
T_punto = compose_matrix(angles=orient, translate=trans)
T_punto_bot = np.matmul(T_bot_inv, T_punto)
punto_x = T_punto_bot[0][3]
punto_y = T_punto_bot[1][3]
print("goal from bot:", punto_x, punto_y)
giro = atan2(punto_y, punto_x)
print("angulo_a_girar: ", degrees(giro))
print("\n")
if ((sqrt((pose.position.x - x_goal)**2 +
(pose.position.y - y_goal)**2)) < 0.01):
twist.linear.x = 0
twist.angular.z = 0
else:
twist.angular.z = k_w * (giro)
twist.linear.x = k_v * sqrt((pose.position.x - x_goal)**2 +
(pose.position.y - y_goal)**2)
if (twist.linear.x > 0.3): #control velocidad
twist.linear.x = 0.3
if (twist.angular.z > 1): #control velocidad angular
twist.angular.z = 1
if (twist.angular.z < (-1)): #control velocidad angular
twist.angular.z = 1 * -1
twist_pub.publish(twist)
print(twist)
rate.sleep()
def controller():
rospy.sleep(1)
print("en el controler")
k_v=1;k_w=1;
# path=[[-0.6,0],[-3.5,0],[-3.5,3.5],[1.5,3.5],[1.5,-1.5],[3.5,-1.5],[3.5,-8.0],[-2.5,-8.0],[-2.5,-5.5],[1.5,-5.5],[1.5,-3.5],[-1,-3.5]]
# x_goal=path[0][0];y_goal=path[0][1];
x_goal=traj.poses.pop(0).pose.position.x;y_goal=path[0][1];
next=0
while not rospy.is_shutdown():
print("true goal:", x_goal, y_goal)
(roll_bot,pitch_bot,yaw_bot)=euler_from_quaternion([pose.orientation.x,pose.orientation.y,pose.orientation.z,pose.orientation.w])
trans = np.array([pose.position.x,pose.position.y,0])
orient = np.array([roll_bot,pitch_bot,yaw_bot])
T_bot = compose_matrix(angles=orient, translate=trans)
T_bot_inv=np.linalg.inv(T_bot)
trans = np.array([x_goal,y_goal,0])
orient = np.array([0,0,0])
T_punto = compose_matrix(angles=orient, translate=trans)
T_punto_bot=np.matmul(T_bot_inv,T_punto)
punto_x=T_punto_bot[0][3];punto_y=T_punto_bot[1][3]
# print("goal from bot:", punto_x, punto_y)
giro=atan2(punto_y,punto_x)
# print("angulo_a_girar: ", degrees(giro))
# print("\n")
if ((sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2))<0.5):
if (not next>=len(path)):
x_goal=path[next][0];y_goal=path[next][1];
# x_goal=traj.poses.pose.point.x;y_goal=traj.poses.pose.point.y
next=next+1
else:
if ((sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2))<0.05):
giro=0;pose.position.x=x_goal;pose.position.y=y_goal;
twist.angular.z=k_w*(giro)
twist.linear.x=k_v*sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2)
if(twist.linear.x>0.3): #control velocidad lineal
twist.linear.x=0.3
if(twist.angular.z>radians(60)): #control velocidad angular
twist.angular.z=radians(60)
if(twist.angular.z<(radians(-60))): #control velocidad angular
twist.angular.z=radians(-60)
twist_pub.publish(twist)
# print(twist)
rate.sleep()
def controller(self):
rospy.sleep(1)
print("en el controler")
k_v=1;k_w=1;
punto=self.traj.poses.pop(0).pose;
x_goal=punto.position.x
y_goal=punto.position.y
self.new=False
while not rospy.is_shutdown() and not self.new:
print("true goal:", x_goal, y_goal)
(roll_bot,pitch_bot,yaw_bot)=euler_from_quaternion([pose.orientation.x,pose.orientation.y,pose.orientation.z,pose.orientation.w])
trans = np.array([pose.position.x,pose.position.y,0])
orient = np.array([roll_bot,pitch_bot,yaw_bot])
T_bot = compose_matrix(angles=orient, translate=trans)
T_bot_inv=np.linalg.inv(T_bot)
trans = np.array([x_goal,y_goal,0])
orient = np.array([0,0,0])
T_punto = compose_matrix(angles=orient, translate=trans)
T_punto_bot=np.matmul(T_bot_inv,T_punto)
punto_x=T_punto_bot[0][3];punto_y=T_punto_bot[1][3]
print("goal from bot:", punto_x, punto_y)
giro=atan2(punto_y,punto_x)
print("angulo_a_girar: ", degrees(giro))
# print("\n")
if ((sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2))<0.5):
if (len(self.traj.poses)>0):
punto=self.traj.poses.pop(0).pose;
x_goal=punto.position.x
y_goal=punto.position.y
else:
if ((sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2))<0.01):
giro=0;pose.position.x=x_goal;pose.position.y=y_goal;
twist.angular.z=k_w*(giro)
twist.linear.x=k_v*sqrt((pose.position.x-x_goal)**2+(pose.position.y-y_goal)**2)
if(twist.linear.x>0.3): #control velocidad lineal
twist.linear.x=0.3
if(twist.angular.z>radians(60)): #control velocidad angular
twist.angular.z=radians(60)
if(twist.angular.z<(radians(-60))): #control velocidad angular
twist.angular.z=radians(-60)
twist_pub.publish(twist)
print(twist)
rate.sleep()
def get_kinematic_chain(self, target, source, reference):
"""
Gets the pose for all joints in the kinematic chain from source to
target.
:param target: name of target joint
:param source : name of source joint
:param reference: name of reference joint as needed by ros function
listener.chain
:type target: string
:type source: string
:type reference: string
:return: kinematic chain, transforms as translation and quaternions,
transforms as matrices
:rtype: list of strings, list of transforms, list of numpy matrices
"""
# All frames here.
listener = self.listener
chain = listener.chain(target, rospy.Time(), source, rospy.Time(),
reference)
Ts, TMs = [], []
np.set_printoptions(precision=4, suppress=True)
TMcum = np.eye(4, 4)
for i in range(len(chain) - 1):
t = listener.lookupTransform(chain[i + 1], chain[i], rospy.Time())
t = [[np.float64(_) for _ in t[0]], [np.float64(_) for _ in t[1]]]
t1_euler = tfx.euler_from_quaternion(t[1])
tm = tfx.compose_matrix(translate=t[0], angles=t1_euler)
Ts.append(t)
TMs.append(tm)
t = listener.lookupTransform(chain[i + 1], chain[0], rospy.Time())
t1_euler = tfx.euler_from_quaternion(t[1])
tm = tfx.compose_matrix(translate=t[0], angles=t1_euler)
TMcum = np.dot(TMs[i], TMcum)
eye = np.dot(tm, np.linalg.inv(TMcum))
assert (np.allclose(eye - np.eye(4, 4), np.zeros((4, 4)),
atol=0.1))
# Sanity check to make sure we understand what is happening here.
# for i in range(len(chain)-1):
# t = listener.lookupTransform(chain[i+1], chain[0], rospy.Time(0))
# tm = tfx.compose_matrix(translate=t[0], angles=t[1])
# TMcum = np.dot(TMs[i], TMcum)
# eye = np.dot(tm, np.linalg.inv(TMcum))
# print(eye-np.eye(4,4))
# assert(np.allclose(eye-np.eye(4,4), np.zeros((4,4)), atol=1e-2))
return chain, Ts, TMs
def parseUrdfJoint(xml, jointNames, jonintTree):
'''获取每个关节的实际位姿'''
matrix = OrderedDict()
result = OrderedDict()
qmatrix = OrderedDict()
# axises=[]
#获取关节参数
for jname in jointNames:
jointNode = xml.xpath("//joint[@name='%s']" % jname)[0]
originNode = jointNode.find("origin")
axisNode = jointNode.find("axis")
origin_xyz = [0.0, 0.0, 0.0] if originNode is None else map(
float, originNode.attrib['xyz'].split())
origin_rpy = [0.0, 0.0, 0.0] if originNode is None else map(
float, originNode.attrib['rpy'].split())
# axis_xyz=[0.0,0.0,1.0] if axisNode is None else map(float,axisNode.attrib['xyz'].split())
mat = compose_matrix(angles=origin_rpy, translate=origin_xyz)
pname = getParentJointName(jonintTree, jname) #获取当前关节的父关节名称
if pname is None:
matrix[jname] = mat
else:
matrix[jname] = matrix[pname].dot(mat)
scale, shear, angles, translate, perspective = decompose_matrix(
matrix[jname])
result[jname] = [
np.array(translate).round(8),
np.array(angles).round(8)
]
qmatrix[jname] = [
translate.round(8),
quaternion_from_euler(*angles).round(8)
]
# axises.append(axis_xyz)
return result
def __init__(self, robot, remove_object_model_func):
Task2State.__init__(
self,
robot,
outcomes=['finish', 'continue'],
input_keys=['object_count', 'orig_channel_xy_yaw'],
output_keys=['object_count', 'orig_channel_xy_yaw'])
self.remove_object_model_func = remove_object_model_func
self.global_place_channel_z = rospy.get_param(
'~global_place_channel_z')
self.global_place_channel_yaw = rospy.get_param(
'~global_place_channel_yaw')
self.place_z_margin = rospy.get_param('~place_z_margin')
self.place_z_offset = rospy.get_param('~place_z_offset')
self.object_num = rospy.get_param('~object_num')
self.skip_ungrasp = rospy.get_param('~skip_ungrasp', False)
self.do_channel_recognition = rospy.get_param(
'~do_channel_recognition')
self.single_object_mode = rospy.get_param('~single_object_mode')
grasping_point = rospy.get_param('~grasping_point')
self.grasping_yaw = rospy.get_param('~grasping_yaw')
self.grasping_point_coords = tft.compose_matrix(
translate=[
grasping_point[0], grasping_point[1], grasping_point[2]
],
angles=[0, 0, self.grasping_yaw])
def trajectory(self, node: str) -> List[Dict[str, List]]:
# collect all timed nodes corresponding to the node to track
traj = []
for nname, ndata in self._graph.nodes(data=True):
if ndata['__name__'] == node:
# Filter out Duckiebot nodes that are not connected to
# another Duckiebot node
dbot_type = AutolabReferenceFrame.TYPE_DUCKIEBOT_FOOTPRINT
if ndata["type"] == dbot_type:
if not self._graph.has_neighbor_of_type(nname, dbot_type):
continue
T = tr.compose_matrix(translate=ndata["pose"].t,
angles=tr.euler_from_quaternion(
ndata["pose"].q))
traj.append({
'timestamp':
ndata["time"],
'pose':
T.tolist(),
# TODO: this is a hack, should be done properly
'observers':
list({
p[0]
for e in self._graph.out_edges(nname)
for p in self._graph.in_edges(e[1])
} if 'footprint' in nname else {})
})
# sort trajectory by time
strategy = lambda tf: tf['timestamp']
traj = sorted(traj, key=strategy)
# ---
return traj
def load_config(self):
anchor_data = yaml.load(file(rospkg.RosPack().get_path('odom_uwb')+"/config/"+"husky.yaml",'r')) # Need RosPack get_path to find the file path
trans = anchor_data['trans']
rot = anchor_data['rot']
return tr.compose_matrix(angles = rot, translate = trans)
def filterPointCloud(bin_num, pts, source_pc2_kinect_header, height=None, width=None, retPointCloud=True):
# input height/width if you need filtered uv
with Timer('b1'):
global tfListener
source_pc_kinect = PointCloud()
source_pc_kinect.header = source_pc2_kinect_header
with Timer('b2'):
source_pc_kinect.points = pts
# transform point cloud from kinect frmae to shelf frame
with Timer('b3'):
import tf.transformations as tfm
import numpy as np
(pos, ori) = tfListener.lookupTransform('/shelf', source_pc_kinect.header.frame_id, source_pc_kinect.header.stamp)
points = []
T = np.dot(tfm.compose_matrix(translate=pos) , tfm.quaternion_matrix(ori) )
for pt in pts:
#tmp = np.dot(T, np.array(pt+[1])).tolist()
points.append( np.dot(T, np.array(pt+[1])).tolist() )
pts = []
cnt = 0
with Timer('b4'):
if height and width:
filtered_uvs = []
filtered_uvs_mask = [False for i in xrange(height*width)]
for point in points: # change source to source_pc_kinect
if ~math.isnan(point[0]) and inside_bin(point, bin_num):
pts.append(point) # can be removed becuase we only need mask
filtered_uvs_mask[cnt] = True
filtered_uvs.append([cnt//width, cnt%width])
cnt += 1
else:
for point in points: # change source to source_pc_kinect
if inside_bin(point, bin_num):
pts.append(point)
if retPointCloud:
with Timer('b5'):
filtered_pc_shelf = PointCloud()
filtered_pc_shelf.header = source_pc2_kinect_header
filtered_pc_shelf.header.frame_id = '/shelf'
filtered_pc_shelf.points = [Point32(pt[0], pt[1], pt[2]) for pt in pts]
# render filtered point cloud
filtered_pc_map = tfListener.transformPointCloud('/map', filtered_pc_shelf)
createPointsMarker(filtered_pc_map.points, marker_id=2, frame_id='/map', rgba=(0,1,0,1)) #points are in Point32
with Timer('b6'):
if height and width:
if retPointCloud:
return (filtered_pc_map, filtered_uvs, filtered_uvs_mask)
else:
return filtered_uvs_mask
else:
return filtered_pc_map
def clickedCB(data):
global askedMarkers
global listener
global worldTransform
global measuredMarkers
global clickNum
print 'clicked'
try:
(tipTrans, tipRot) = listener.lookupTransform('/mocha_world', '/tip',
rospy.Time(0))
measuredMarkers[clickNum, :] = tipTrans
clickNum = clickNum + 1
if clickNum == 8:
clickNum = 0
print 'finished clicking'
print 'AskedMarkers:'
print askedMarkers
print 'measured markers:'
print measuredMarkers
(worldRotupdate,
worldTransUpdate) = rigid_transform_3D(measuredMarkers,
askedMarkers)
worldRot4 = np.identity(4)
for i in range(3):
worldRot4[(0, 1, 2), i] = worldRotupdate[:, i]
worldTransformUpdate = TR.compose_matrix(
translate=worldTransUpdate,
angles=TR.euler_from_matrix(worldRot4))
worldTransformationUpdateInv = TR.inverse_matrix(
worldTransformUpdate)
worldTransform = np.dot(worldTransformationUpdateInv,
worldTransform)
except ():
nothing = 0
def __init__(self, robot, add_object_model_func):
Task2State.__init__(self,
robot,
outcomes=['succeeded', 'failed'],
input_keys=['object_count'],
output_keys=['object_count'])
self.add_object_model_func = add_object_model_func
self.do_object_recognition = rospy.get_param('~do_object_recognition')
self.joint_torque_thresh = rospy.get_param('~joint_torque_thresh')
self.skip_grasp = rospy.get_param('~skip_grasp', False)
self.object_grasping_height = rospy.get_param(
'~object_grasping_height')
self.object_interval = rospy.get_param('~object_interval')
self.lane_interval = rospy.get_param('~lane_interval')
self.global_first_object_pos = rospy.get_param(
'~global_first_object_pos')
self.global_object_yaw = rospy.get_param('~global_object_yaw')
self.grasp_land_mode = rospy.get_param('~grasp_land_mode')
self.reset_realsense_odom = rospy.get_param('~reset_realsense_odom')
self.stop_if_grasp_failed = rospy.get_param('~stop_if_grasp_failed')
self.reset_realsense_client = rospy.ServiceProxy(
'/realsense1/odom/reset', std_srvs.srv.Empty)
grasping_point = rospy.get_param('~grasping_point')
grasping_yaw = rospy.get_param('~grasping_yaw')
self.grasping_point_coords = tft.compose_matrix(
translate=[
grasping_point[0], grasping_point[1], grasping_point[2]
],
angles=[0, 0, grasping_yaw])
self.grasp_force_landing_mode = rospy.get_param(
'~grasp_force_landing_mode')
self.force_landing_height = rospy.get_param('~force_landing_height')
def ref_pose_cb(self, some_pose): # Takes target pose, returns ref se3
rospy.logdebug("ref_pose_cb called in velocity_control.py")
p = np.array([some_pose.position.x, some_pose.position.y, some_pose.position.z])
quat = np.array([some_pose.orientation.x, some_pose.orientation.y, some_pose.orientation.z, some_pose.orientation.w])
goal_tmp = tr.compose_matrix(angles=tr.euler_from_quaternion(quat, 'sxyz'), translate=p)
with self.mutex:
self.T_goal = goal_tmp
def __init__(self, robot):
Task2State.__init__(
self,
robot,
outcomes=['succeeded', 'failed', 'adjust_again'],
input_keys=['orig_global_trans', 'search_count', 'search_failed'],
output_keys=['orig_global_trans', 'search_count', 'search_failed'])
self.skip_adjust_grasp_position = rospy.get_param(
'~skip_adjust_grasp_position', False)
self.do_object_recognition = rospy.get_param('~do_object_recognition')
self.object_yaw_thresh = rospy.get_param('~object_yaw_thresh')
self.global_object_yaw = rospy.get_param('~global_object_yaw')
self.grasping_yaw = rospy.get_param('~grasping_yaw')
self.recognition_wait = rospy.get_param('~recognition_wait')
self.adjust_grasp_image_type = rospy.get_param(
'~adjust_grasp_image_type', 'depth')
self.object_pos_thresh = rospy.get_param('~object_pos_thresh')
grasping_point = rospy.get_param('~grasping_point')
grasping_yaw = rospy.get_param('~grasping_yaw')
self.grasping_point_coords = tft.compose_matrix(
translate=[
grasping_point[0], grasping_point[1], grasping_point[2]
],
angles=[0, 0, grasping_yaw])
self.object_pose_sub = rospy.Subscriber(
'rectangle_detection_' + self.adjust_grasp_image_type +
'/target_object_' + self.adjust_grasp_image_type, PoseArray,
self.objectPoseCallback)
self.object_pose = PoseArray()
def read_pos(t, link_states, recorder_pos, joint_states, readLinkState):
#pub_foot_z, pub_body_z):
## Initialization
from geometry_msgs.msg import Wrench, Vector3
from std_msgs.msg import Float64
import rospy
import math
from numpy.linalg import norm
from gazebo_msgs.srv import GetLinkState
from tf.transformations import euler_from_quaternion, quaternion_from_euler, compose_matrix, quaternion_matrix
if readLinkState.value is None:
readLinkState.value = rospy.ServiceProxy("/gazebo/get_link_state", GetLinkState)
## set time from which recording of data (for CSV) should start
record_time = 0.0
# get current angle
phi1 = joint_states.value.position[1]
phi3 = -joint_states.value.position[3]
## definitions according to matlab script/ RAL paper, defined solely for comparison
offset_q1 = -40*math.pi/180
offset_q2 = 40*math.pi/180
q1 = phi1+offset_q1
q3 = phi3+offset_q2
body_pos = readLinkState.value('base_link', 'link')
body_z = body_pos.link_state.pose.position.z
body_x = body_pos.link_state.pose.position.y
lower_leg_pos = readLinkState.value('link2_link', 'link')
(knee_x, knee_y, knee_z) = [lower_leg_pos.link_state.pose.position.x, lower_leg_pos.link_state.pose.position.y, lower_leg_pos.link_state.pose.position.z]
orientation_q = lower_leg_pos.link_state.pose.orientation
orientation_list = [orientation_q.x, orientation_q.y, orientation_q.z, orientation_q.w]
#matrix_k = quaternion_matrix(orientation_list)
#matrix_k[:3,3] = [knee_x, knee_y, knee_z]
(roll, pitch, yaw) = euler_from_quaternion (orientation_list)
matrix_knee = compose_matrix(scale=None, shear=None, angles=[roll, pitch, yaw], translate=[knee_x, knee_y, knee_z], perspective=None)
matrix_kf = compose_matrix(scale=None, shear=None, angles=[0, 0, 0], translate=[0, 0.08, 0], perspective=None)
matrix_foot = np.dot(matrix_knee, matrix_kf)
foot_z = matrix_foot[2][3]#-0.0239
if t > record_time:
recorder_pos.record_entry(t, body_x, body_z, foot_z, q1, q3, phi1, phi3)
def se3_to_T(rvec, tvec):
"""Convert an se(3) OpenCV pose to a 4x4 transformation matrix.
OpenCV poses are an Euler-Rodrigues rotation vector and a translation vector.
"""
R, _ = cv2.Rodrigues(rvec)
euler_angles = transformations.euler_from_matrix(R)
return transformations.compose_matrix(angles=euler_angles, translate=np.squeeze(tvec))
def ref_poseCB(self, goal_pose): # Takes target pose, returns ref se3
rospy.logdebug("ref_pose_cb called in velocity_control.py")
# p = np.array([some_pose.position.x, some_pose.position.y, some_pose.position.z])
p = np.array([goal_pose.position.x, goal_pose.position.y + 0.04, goal_pose.position.z])
quat = np.array([goal_pose.orientation.x, goal_pose.orientation.y, goal_pose.orientation.z, goal_pose.orientation.w])
goal_tmp = tr.compose_matrix(angles=tr.euler_from_quaternion(quat, 'sxyz'), translate=p) # frame is spatial 'sxyz', return Euler angle from quaternion for specified axis sequence.
with self.mutex:
self.original_goal = goal_tmp
def tf_to_T(tf):
"""Convert a (tx, ty, tz, rx, ry, rz) tuple to a 4x4 transformation matrix."""
if len(tf) == 6:
# Single transform
return transformations.compose_matrix(angles=tf[3:], translate=tf[:3])
else:
# Array of transforms
assert tf.shape[1] == 6, "Transform array must be Nx6"
return np.vstack([tf_to_T(row) for row in tf])
def evaluate(t_rc, camera_rb_poses, all_image_coords, all_calibration_markers,
t_co, camera_mat, distortion_coeffs):
# Get the optical frame pose in the mocap frame
t_rc = transformations.compose_matrix(translate=t_rc[:3],
angles=(t_rc[3:]))
t_mos = calculate_t_mos(camera_rb_poses, t_rc, t_co)
shape = (8, 5)
square_size = 0.035
marker_z = 0.012
marker_positions = [
(-square_size, 0, -marker_z),
(square_size * 8, 0, -marker_z),
(square_size * 8, square_size * 5, -marker_z),
(-square_size, square_size * 5, -marker_z),
]
total_error = 0
for t_mo, image_coords, calibration_markers in zip(
t_mos, all_image_coords, all_calibration_markers):
checkerboard_coords = np.zeros((shape[0] * shape[1], 3), np.float32)
checkerboard_coords[:, :2] = (
np.mgrid[0:shape[0], 0:shape[1]].T.reshape(-1, 2)) * square_size
retval, rvec, tvec = cv2.solvePnP(checkerboard_coords, image_coords,
camera_mat, distortion_coeffs)
t_o_ch = calibration_common.se3_to_T(rvec, tvec)
t_m_ch = np.matmul(t_mo, t_o_ch)
total_distance = 0
for marker in marker_positions:
marker = np.matmul(t_m_ch, np.append(marker, 1))[:3]
match, distance = find_closest_marker_distance(
marker, calibration_markers)
print(match[2] - marker[2])
total_distance += distance
total_distance /= 4
print("*" * 80)
# To handle when the checkerboard was flipped during calibration
checkerboard_coords = checkerboard_coords[::-1]
retval, rvec, tvec = cv2.solvePnP(checkerboard_coords, image_coords,
camera_mat, distortion_coeffs)
t_o_ch = calibration_common.se3_to_T(rvec, tvec)
t_m_ch = np.matmul(t_mo, t_o_ch)
other_total_distance = 0
for marker in marker_positions:
marker = np.matmul(t_m_ch, np.append(marker, 1))[:3]
match, distance = find_closest_marker_distance(
marker, calibration_markers)
other_total_distance += distance
print(match[2] - marker[2])
other_total_distance /= 4
if other_total_distance < total_distance:
total_distance = other_total_distance
total_error += total_distance
return total_error / len(t_mos)
def _inverse_tuples(t):
trans, rot = t
euler = tr.euler_from_quaternion(rot)
m = tr.compose_matrix(translate=trans, angles=euler)
m_inv = tr.inverse_matrix(m)
trans_inv = tr.translation_from_matrix(m_inv)
rot_inv = tr.rotation_matrix(*tr.rotation_from_matrix(m_inv))
quat_inv = tr.quaternion_from_matrix(rot_inv)
return (trans_inv, quat_inv)
def matrix_from_xyzquat_np_array(arg1, arg2=None):
if arg2 is not None:
translate = arg1
quaternion = arg2
else:
translate = arg1[0:3]
quaternion = arg1[3:7]
return np.dot(compose_matrix(translate=translate),quaternion_matrix(quaternion))
def kinect_to_base(self,kinect_pose):
try:
time.sleep(1)
(self.transl,self.quat)=self.listener.lookupTransform('base','kinect',rospy.Time(0))
self.rot = euler_from_quaternion(self.quat)
self.tf_SE3 = compose_matrix(angles=self.rot,translate = self.transl)
print self.tf_SE3
base_pos=numpy.dot(self.tf_SE3,kinect_pose)
print base_pos
# print self.cmd_pos
except (tf.Exception):
rospy.logerr("Could not transform from "\
"{0:s} to {1:s}".format(base,kinect))
pass
return base_pos
def animate_2dsynced(data, shape_id, figfname):
fig, ax = plt.subplots()
fig.set_size_inches((7,7))
probe_radius = 0.004745 # probe1: 0.00626/2 probe2: 0.004745
sub = 1 # subsample rate
tip_pose = data['tip_poses_2d']
object_pose = data['object_poses_2d']
force = data['force_2d']
patches = []
# add the object as polygon
shape_db = ShapeDB()
shape_polygon = shape_db.shape_db[shape_id]['shape'] # shape of the objects presented as polygon.
shape_polygon_3d = np.hstack((np.array(shape_polygon), np.zeros((len(shape_polygon), 1)), np.ones((len(shape_polygon), 1))))
print 'object_pose', len(object_pose), 'tip_pose', len(tip_pose), 'force', len(force)
plt.ion()
for i in (range(0, len(tip_pose), sub)):
plt.cla()
T = tfm.compose_matrix(translate = object_pose[i][0:2] + [0], angles = (0,0,object_pose[i][2]) )
shape_polygon_3d_world = np.dot(T, shape_polygon_3d.T)
obj = mpatches.Polygon(shape_polygon_3d_world.T[:,0:2], closed=True, color='blue', alpha=0.05)
ax.add_patch(obj)
# add the probes as circle
circle = mpatches.Circle(tip_pose[i][0:2], probe_radius, ec="none", color='red', alpha=0.5)
ax.add_patch(circle)
# add the force
ax.arrow(tip_pose[i][0], tip_pose[i][1], force[i][0]/100, force[i][1]/100, head_width=0.005, head_length=0.01, fc='k', ec='k')
#arrow = mpatches.Arrow(tip_pose[i][0], tip_pose[i][1], force[i][0],
# force[i][1], head_width=0.05, head_length=0.1, fc='k', ec='k')
#ax.add_patch(arrow)
# render it
plt.axis([-0.3, 0.3, -0.3, 0.3])
#plt.axis('equal')
plt.draw()
#time.sleep(0.1)
plt.show()
def poseTomatrix(pose):
"""Converts a Pose message to a matrix 4.
"""
assert isinstance(pose, Pose)
T = ((pose.position.x,
pose.position.y,
pose.position.z, )
)
R = transformations.euler_from_quaternion((
pose.orientation.x,
pose.orientation.y,
pose.orientation.z,
pose.orientation.w)
)
return transformations.compose_matrix(translate=T,
angles=R)
def goal_from_tag_pose(self, tag_pose):
tf_mat = tft.compose_matrix(angles=self.grasp_transform[3:])
tag_q = (tag_pose.pose.orientation.x, tag_pose.pose.orientation.y,
tag_pose.pose.orientation.z, tag_pose.pose.orientation.w)
tag_mat = tft.quaternion_matrix(tag_q)
new_mat = np.dot(tag_mat, tf_mat)
dxyz = self.grasp_transform[:3]
dxyz.append(1)
d_xyz_base = np.matrix(dxyz).T
d_xyz_tag = np.dot(tag_mat, d_xyz_base).A1[:3].tolist()
new_q = tft.quaternion_from_matrix(new_mat)
grasp_pose = PoseStamped()
grasp_pose.header.stamp = rospy.Time.now()
grasp_pose.header.frame_id = tag_pose.header.frame_id
grasp_pose.pose.position.x = tag_pose.pose.position.x + d_xyz_tag[0]
grasp_pose.pose.position.y = tag_pose.pose.position.y + d_xyz_tag[1]
grasp_pose.pose.position.z = tag_pose.pose.position.z + d_xyz_tag[2]
grasp_pose.pose.orientation = Quaternion(*new_q)
self.test_pub_2.publish(grasp_pose)
return grasp_pose
def tf_base_to_stylus (self):
try:
(self.transl,self.quat) = self.tfListener.lookupTransform('base','stylus',rospy.Time(0))
#lookupTransrom is a method which returns the transfrom between two coordinate frames.
#lookupTransfrom returns a translation and a quaternion
self.rot= euler_from_quaternion(self.quat) #Get euler angles from the quaternion
self.tf_SE3 = compose_matrix(angles=self.rot,translate=self.transl)
#Store the transformation in a format compatible with gemoetr_msgs/Twist
self.transf = Twist(Vector3(self.transl[0],self.transl[1],self.transl[2]),(Vector3(self.rot[0],self.rot[1],self.rot[2])))
#Publish the transformation
self.Tsb.publish(self.transf)
except (tf.LookupException, tf.ConnectivityException, tf.ExtrapolationException):
self.tf_SE3 = None
pass
#If exceptions occur, skip trying to lookup the transform.
#It is encouraged to publish a logging message to rosout with rospy.
#i.e:
#rospy.logerr("Could not transform from %s to %s,"base","stylus")
return
def params_to_matrix(self, params):
# uses euler angles.... assumes we aren't near singularities
translate = params[0:3]
angles = params[3:6]
return transformations.compose_matrix(translate=translate, angles=angles)
def matrix_from_xyzrpy(translate, rpy):
return np.dot(tfm.compose_matrix(translate=translate) ,
tfm.euler_matrix(*rpy)).tolist()
except (tf.LookupException, tf.ConnectivityException, tf.ExtrapolationException, tf.Exception) as e:
rospy.logwarn("Could not find transformation: " + str(e))
continue
# rospy.loginfo("Got the matrix:\n%s", transformation_matrix)
translation = transformation_matrix[:3, 3]
rotation = transformation_matrix[:3, :3]
permutation_matrix = numpy.array([[0, 0., -1],
[-1, 0, 0],
[0, 1, 0]])
new_rotation = permutation_matrix.dot(rotation)
R = numpy.eye(4)
R[:3, :3] = new_rotation
T1 = transformations.compose_matrix(translate= -translation)
T2 = transformations.compose_matrix(translate=translation)
# full_transformation_matrix = T1.dot(R).dot(T2)
full_transformation_matrix = T2.dot(R).dot(T1)
# rospy.loginfo("Full transformation matrix:\n%s", full_transformation_matrix)
resulting_matrix = full_transformation_matrix.dot(transformation_matrix)
# rospy.loginfo("The result is:\n%s", resulting_matrix)
full_translation = transformations.translation_from_matrix(resulting_matrix)
full_rotation_q = transformations.quaternion_from_matrix(resulting_matrix)
br.sendTransform(full_translation, full_rotation_q, hdr.stamp, "flipped_frame", parent_frame)
def plan(self):
# plan() return an object includes attributes
# q_traj,
# snopt_info_iktraj,
# infeasible_constraint_iktraj
# snopt_info_ik
# 1. Get a handle for the service
# todo: makes warning if ikTrajServer does not exist
ikTrajServer_pub = rospy.Publisher('/ikTrajServer', std_msgs.msg.String, queue_size=100)
argin = {}
# 2. prepare input arguments
if self.q0 is not None:
argin['q0'] = self.q0
else:
# get robot joints from published topic # sensor_msgs/JointState
while True:
APCrobotjoints = ROS_Wait_For_Msg(self.joint_topic, sensor_msgs.msg.JointState).getmsg()
q0 = APCrobotjoints.position
if len(q0) < 6:
continue
else:
argin['q0'] = q0
break
# 2.1 transform tip pose to hand pose
if self.target_tip_pos is not None:
tip_hand_tfm_mat = np.dot( tfm.compose_matrix(translate=self.tip_hand_transform[0:3]),
tfm.compose_matrix(angles=self.tip_hand_transform[3:6]) )
tip_world_rot_mat = tfm.quaternion_matrix(self.target_tip_ori)
tip_world_tfm_mat = np.dot(tfm.compose_matrix(translate=self.target_tip_pos) , tip_world_rot_mat)
hand_tip_tfm_mat = np.linalg.inv(tip_hand_tfm_mat)
hand_world_tfm_mat = np.dot(tip_world_tfm_mat, hand_tip_tfm_mat)
target_hand_pos = tfm.translation_from_matrix(hand_world_tfm_mat)
argin['target_hand_pos'] = target_hand_pos.tolist()
if self.target_tip_ori is not None:
tip_hand_tfm_mat = tfm.compose_matrix(angles=self.tip_hand_transform[3:6])
tip_world_rot_mat = tfm.quaternion_matrix(self.target_tip_ori)
hand_tip_tfm_mat = np.linalg.inv(tip_hand_tfm_mat)
hand_world_tfm_mat = np.dot(tip_world_tfm_mat, hand_tip_tfm_mat)
target_hand_ori = tfm.quaternion_from_matrix(hand_world_tfm_mat)
argin['target_hand_ori'] = roshelper.ros2matlabQuat(target_hand_ori.tolist()) # tolist: so that its serializable
# 2.2 prepare other options
argin['straightness'] = self.straightness
argin['target_link'] = self.target_link
argin['pos_tol'] = self.pos_tol
argin['ori_tol'] = self.ori_tol
if self.inframebb is not None:
argin['inframebb'] = self.inframebb
if self.target_joint_bb is not None:
argin['target_joint_bb'] = self.target_joint_bb
argin['N'] = self.N
argin['tip_hand_transform'] = self.tip_hand_transform
if self.ik_only:
argin['ik_only'] = 1
else:
argin['ik_only'] = 0
argin_json = json.dumps(argin) # convert to json format and
for i in range(10):
# 3. call it
#print 'calling matlab drake...'
argin_json_ = argin_json + '\n'
try:
tn = telnetlib.Telnet(self.ikServerAddress[0], self.ikServerAddress[1])
time.sleep(0.05)
tn.write(argin_json_)
except socket.error as inst:
print '[Drake] Matlab server not up now.'
if i==9:
return None
continue
time.sleep(0.05)
# 4. parse output arguments
try:
ret_json = tn.read_all()
ret = json.loads(ret_json) # convert from json to struct
break
except:
print '[Drake] read_all() failed, try again.'
if i==9:
return None
continue
return Plan(ret)
def matrix_from_xyzquat(translate, quaternion):
return np.dot(tfm.compose_matrix(translate=translate) ,
tfm.quaternion_matrix(quaternion)).tolist()