def do_euclidian_clustering(cloud):
# Euclidean Clustering
white_cloud = pcl_helper.XYZRGB_to_XYZ(cloud) # <type 'pcl._pcl.PointCloud'>
tree = white_cloud.make_kdtree() # <type 'pcl._pcl.KdTree'>
ec = white_cloud.make_EuclideanClusterExtraction()
ec.set_ClusterTolerance(0.02) # for hammer
ec.set_MinClusterSize(10)
ec.set_MaxClusterSize(250)
ec.set_SearchMethod(tree)
cluster_indices = ec.Extract() # indices for each cluster (a list of lists)
# Assign a color to each cluster
cluster_color = pcl_helper.random_color_gen()
#cluster_color = pcl_helper.get_color_list(len(cluster_indices))
color_cluster_point_list = []
for j, indices in enumerate(cluster_indices):
for i, indice in enumerate(indices):
color_cluster_point_list.append([white_cloud[indice][0], white_cloud[indice][1], white_cloud[indice][2], pcl_helper.rgb_to_float(cluster_color)])
# Create new cloud containing all clusters, each with unique color
cluster_cloud = pcl.PointCloud_PointXYZRGB()
cluster_cloud.from_list(color_cluster_point_list)
# publish to cloud
ros_cluster_cloud = pcl_helper.pcl_to_ros(cluster_cloud)
# save to local
pcl.save(cluster_cloud, 'cluster.pcd')
return cluster_cloud
def do_euclidian_clustering(cloud):
# Euclidean Clustering
white_cloud = pcl_helper.XYZRGB_to_XYZ(cloud)
tree = white_cloud.make_kdtree()
ec = white_cloud.make_EuclideanClusterExtraction()
ec.set_ClusterTolerance(5)
ec.set_MinClusterSize(1)
ec.set_MaxClusterSize(3500)
ec.set_SearchMethod(tree)
cluster_indices = ec.Extract()
cluster_color = pcl_helper.random_color_gen()
#cluster_color = pcl_helper.get_color_list(len(cluster_indices))
color_cluster_point_list = []
for j, indices in enumerate(cluster_indices):
for i, indice in enumerate(indices):
color_cluster_point_list.append([
white_cloud[indice][0], white_cloud[indice][1],
white_cloud[indice][2],
pcl_helper.rgb_to_float(cluster_color)
])
cluster_cloud = pcl.PointCloud_PointXYZRGB()
cluster_cloud.from_list(color_cluster_point_list)
ros_cluster_cloud = pcl_helper.pcl_to_ros(cluster_cloud)
return cluster_cloud
def callback(input_ros_msg):
cloud = pcl_helper.ros_to_pcl(input_ros_msg)
print("INPUT: ", cloud, type(cloud))
cloud = pcl_helper.XYZRGB_to_XYZ(cloud)
mean_k = 10
thresh = 0.001
cloud = do_statistical_outlier_filtering(cloud, mean_k, thresh)
color = pcl_helper.random_color_gen()
cloud = pcl_helper.XYZ_to_XYZRGB(cloud, color)
cloud_new = pcl_helper.pcl_to_ros(cloud)
print("OUTPUT: ", cloud, type(cloud))
pub.publish(cloud_new)
def Process_rawPCL(pclpcRawIn):
DebugDumpMsgPCL(pclpcRawIn)
pclRecs = [] # For dev/debug display. Container for point cloud records: tuple (pclObj, pclName# )
pclRecs.append((pclpcRawIn, "pclpcRawIn"))
pclRecsDownSampled = pclproc.PCLProc_DownSampleVoxels(pclpcRawIn)
pclRecs += pclRecsDownSampled
pclpcDownSampled, pclpcDownSampledName = pclRecsDownSampled[0]
# PassThrough Filter
pclRecsRansac = pclproc.PCLProc_Ransac(pclpcDownSampled)
pclRecs += pclRecsRansac
# Extract inliers and outliers
pclpcPassZ, pclpcPassZIn, pclpcPassZOut = pclRecsRansac[0][0], pclRecsRansac[1][0], pclRecsRansac[2][0]
pclpcTable, pclpcObjects = pclpcPassZIn, pclpcPassZOut # Rename for clarity
# Euclidean Clustering
pclpObjectsNoColor = pcl_helper.XYZRGB_to_XYZ(pclpcObjects)
clusterIndices, pclpcClusters = pclproc.PCLProc_ExtractClusters(pclpObjectsNoColor)
labelRecs = []
for index, pts_list in enumerate(clusterIndices):
# Get points for a single object in the overall cluster
pcl_cluster = pclpcObjects.extract(pts_list)
msgPCL_cluster = pcl_helper.pcl_to_ros(pcl_cluster) # Needed for histograms... would refactor
# Extract histogram features
chists = pclproc.compute_color_histograms(msgPCL_cluster, doConvertToHSV=True)
normals = get_normals(msgPCL_cluster)
nhists = pclproc.compute_normal_histograms(normals)
feature = np.concatenate((chists, nhists))
# CLASSIFY, retrieve the label for the result
# and add it to detected_objects_labels list
prediction = g_clf.predict(g_scaler.transform(feature.reshape(1, -1)))
label = g_encoder.inverse_transform(prediction)[0]
# Accumulate label records for publishing (and labeling detected objects)
label_pos = list(pclpcObjects[pts_list[0]])
label_pos[2] += 0.3
labelRecs.append((label, label_pos, index))
return labelRecs, pclpcObjects, pclpcTable, pclpcClusters
def detect_objects(cloud_objects, cluster_indices):
# TODO: Classify the clusters!
detected_objects_labels = []
detected_objects = []
labeled_features = []
for index, pts_list in enumerate(cluster_indices):
# TODO: Grab the points for the cluster from the extracted outliers (cloud_objects)
pcl_cluster = cloud_objects.extract(pts_list)
# TODO: Convert the cluster from pcl to ROS using helper function
ros_cluster = pclh.pcl_to_ros(pcl_cluster)
# TODO: Extract histogram features
hists_color = ft.compute_color_histograms(ros_cluster, using_hsv=True)
normals = get_normals(ros_cluster)
hists_normals = ft.compute_normal_histograms(normals)
feature = np.concatenate((hists_color, hists_normals))
#labeled_features.append([feature, model_name])
# TODO: Make the prediction
prediction = clf.predict(scaler.transform(feature.reshape(1, -1)))
# TODO: Retrieve the label for the result
label = encoder.inverse_transform(prediction)[0]
# TODO: Add it to detected_objects_labels list
detected_objects_labels.append(label)
# TODO: Publish a label into RViz
white_cloud = pclh.XYZRGB_to_XYZ(cloud_objects)
label_pos = list(white_cloud[pts_list[0]])
label_pos[2] += .2
object_markers_pub.publish(make_label(label, label_pos, index))
# TODO: Add the detected object to the list of detected objects.
do = DetectedObject()
do.label = label
do.cloud = ros_cluster
detected_objects.append(do)
rospy.loginfo('Detected {} objects: {}'.format(
len(detected_objects_labels), detected_objects_labels))
return detected_objects, detected_objects_labels
def euclidean_clustering(cloud, ClusterTolerance=0.01, MinClusterSize=10, MaxClusterSize=10000):
# Convert XYZRGB point cloud to XYZ since EC uses spatial info only
cloud_xyz = pclh.XYZRGB_to_XYZ(cloud)
# Use k-d tree to decrease the computational burden of searching for neighboring points
tree = cloud_xyz.make_kdtree()
# Create a cluster extraction object
ec = cloud_xyz.make_EuclideanClusterExtraction()
# Set tolerances for distance threshold
# as well as minimum and maximum cluster size (in points)
# NOTE: These are poor choices of clustering parameters
# Your task is to experiment and find values that work for segmenting objects.
ec.set_ClusterTolerance(ClusterTolerance)
ec.set_MinClusterSize(MinClusterSize)
ec.set_MaxClusterSize(MaxClusterSize)
# Search the k-d tree for clusters
ec.set_SearchMethod(tree)
# Extract indices for each of the discovered clusters
cluster_indices = ec.Extract()
return cluster_indices
def PCLProc_Noise(pclpIn):
pclRecs = [
] # For dev/debug display. Container for point cloud records: tuple (pclObj, pclName)
print("pcl.__file__", pcl.__file__)
print("type pcl", type(pclpIn))
if (not type(pclpIn) is pcl._pcl.PointCloud):
pclpIn = pcl_helper.XYZRGB_to_XYZ(pclpIn)
fil = pclpIn.make_statistical_outlier_filter()
numNeighborsToCheck = 20
threshScaleFactor = 1
fil.set_mean_k(numNeighborsToCheck)
fil.set_std_dev_mul_thresh(threshScaleFactor)
pclpNoiseInliers = fil.filter()
fil.set_negative(True)
pclpNoiseOutliers = fil.filter()
pclRecs.append((pclpNoiseInliers, "pclpNoiseInliers"))
pclRecs.append((pclpNoiseOutliers, "pclpNoiseOutliers"))
return (pclRecs)
def ProcessPCL(pclpcRawIn):
pclRecs = [
] # For dev/debug display. Container for point cloud records: tuple (pclObj, pclName# )
pclRecs.append((pclpcRawIn, "pclpcRawIn"))
pclRecsDownSampled = pclproc.PCLProc_DownSampleVoxels(pclpcRawIn)
pclRecs += pclRecsDownSampled
pclpcDownSampled, pclpcDownSampledName = pclRecsDownSampled[0]
# PassThrough Filter
pclRecsRansac = pclproc.PCLProc_Ransac(pclpcDownSampled)
pclRecs += pclRecsRansac
# Extract inliers and outliers
pclpcPassZ, pclpcPassZIn, pclpcPassZOut = pclRecsRansac[0][
0], pclRecsRansac[1][0], pclRecsRansac[2][0]
pclpcTable, pclpcObjects = pclpcPassZIn, pclpcPassZOut
# Euclidean Clustering
pclpObjectsNoColor = pcl_helper.XYZRGB_to_XYZ(pclpcObjects)
pclpcClusters = pclproc.PCLProc_ExtractClusters(pclpObjectsNoColor)
return pclpcObjects, pclpcTable, pclpcClusters
def pcl_callback(ros_pcl_msg):
# Exercise-1 TODOs:
# TODO: Load point cloud data
# TODO: Voxel Grid Downsampling
# TODO: PassThrough Filter
# TODO: RANSAC Plane Segmentation
# TODO: Extract inliers and outliers
# TODO: Save the inliers in .pcd file
# TODO: Run the inliers.pcd file on terminal
# Exercise-2 TODOs:
# TODO: Convert ROS msg to PCL data
# TODO: Voxel Grid Downsampling
# TODO: PassThrough Filter
# TODO: RANSAC Plane Segmentation
# TODO: Extract inliers and outliers
# TODO: Euclidean Clustering
# TODO: Create Cluster-Mask Point Cloud to visualize each cluster separately
# TODO: Convert PCL data to ROS messages
# TODO: Publish ROS messages
# Exercise-3 TODOs:
# TODO: Classify the clusters! (loop through each detected cluster one at a time)
# TODO: Grab the points for the cluster
# TODO: Compute the associated feature vector
# TODO: Make the prediction
# TODO: Publish a label into RViz
# TODO: Add the detected object to the list of detected objects.
# TODO: Publish the list of detected objects
# TODO: Convert ROS msg of type PointCloud2 to PCL data PointXYZRGB format
pcl_data = pclh.ros_to_pcl(ros_pcl_msg)
# TODO: Statistical filter
stat_inlier, stat_outlier = flt.statistical_outlier_filter(pcl_data)
pcl_data = stat_inlier
# TODO: Voxel Grid Downsampling
vox_filt = flt.voxel_downsampling(pcl_data)
# TODO: PassThrough Filter
pass_filt = flt.pass_through_filter(vox_filt)
# TODO: RANSAC Plane Segmentation
# TODO: Extract inliers (objects) and outliers (table)
cloud_table, cloud_objects = flt.ransac(pass_filt)
# TODO: Euclidean Clustering
cluster_indices = flt.euclidean_clustering(cloud_objects)
# TODO: Create Cluster-Mask Point Cloud to visualize each cluster separately
cluster_cloud = flt.visualize_clusters(cloud_objects, cluster_indices)
# TODO: Convert PCL data to ROS messages
ros_table_cloud = pclh.pcl_to_ros(cloud_table)
ros_objects_cloud = pclh.pcl_to_ros(cloud_objects)
ros_cluster_cloud = pclh.pcl_to_ros(cluster_cloud)
# TODO: Publish ROS messages
pcl_table_pub.publish(ros_table_cloud)
pcl_objects_pub.publish(ros_objects_cloud)
pcl_cluster_pub.publish(ros_cluster_cloud)
# TODO: Classify the clusters!
detected_objects_labels = []
detected_objects = []
labeled_features = []
for index, pts_list in enumerate(cluster_indices):
# Grab the points for the cluster from the extracted outliers (cloud_objects)
pcl_cluster = cloud_objects.extract(pts_list)
# TODO: convert the cluster from pcl to ROS using helper function
ros_cluster = pclh.pcl_to_ros(pcl_cluster)
# Extract histogram features
# TODO: complete this step just as is covered in capture_features.py
# Extract histogram features
hists_color = compute_color_histograms(ros_cluster, using_hsv=True)
normals = get_normals(ros_cluster)
hists_normals = compute_normal_histograms(normals)
feature = np.concatenate((hists_color, hists_normals))
#labeled_features.append([feature, model_name])
# Make the prediction, retrieve the label for the result
# and add it to detected_objects_labels list
prediction = clf.predict(scaler.transform(feature.reshape(1,-1)))
label = encoder.inverse_transform(prediction)[0]
detected_objects_labels.append(label)
# Publish a label into RViz
white_cloud = pclh.XYZRGB_to_XYZ(cloud_objects)
label_pos = list(white_cloud[pts_list[0]])
label_pos[2] += .4
object_markers_pub.publish(make_label(label,label_pos, index))
# Add the detected object to the list of detected objects.
do = DetectedObject()
do.label = label
do.cloud = ros_cluster
detected_objects.append(do)
rospy.loginfo('Detected {} objects: {}'.format(len(detected_objects_labels), detected_objects_labels))
# Publish the list of detected objects
# This is the output you'll need to complete the upcoming project!
detected_objects_pub.publish(detected_objects)
def pc2CB(self, msg):
if self.init:
# Convert ROS PointCloud2 msg to PCL data
cloud = pcl_helper.ros_to_pcl(msg)
# Filter the Cloud
fil_cloud = filtering_helper.do_passthrough_filter(
cloud, 'x', self.x_min, self.x_max)
filtered_cloud = filtering_helper.do_passthrough_filter(
fil_cloud, 'y', self.y_min, self.y_max)
# Turn cloud into colorless cloud
colorless_cloud = pcl_helper.XYZRGB_to_XYZ(filtered_cloud)
# Get groups of cluster of points
# clusters = self.getClusters(colorless_cloud, 0.05, 20, 300) # Table on Top AGV - 60~300
clusters = self.getClusters(colorless_cloud, 0.05, 20, 350)
print "len(clusters)", len(clusters)
for i in range(0, len(clusters)):
print "len(clusters[", i, "]", len(clusters[i])
# Get mean points from clusters
mean_pts = []
for cluster in clusters:
mean_pts.append(self.getMeanPoint(cluster, filtered_cloud))
print "len(mean_pts)", len(mean_pts)
for i in range(0, len(mean_pts)):
print "mean_pts[", i, "]", mean_pts[i]
# Remove other points, leave the table points only
table_pts = []
table_pts = self.getTablePoints(mean_pts)
print "len(table_pts)", len(table_pts)
print table_pts
for i in range(len(table_pts)):
self.vizPoint(i, table_pts[i], ColorRGBA(1, 1, 1, 1),
"base_link")
# Get transform of "base_link" relative to "map"
trans = self.getTransform("map", "base_link")
# Finding 2 middle points from 4 table legs and use them as path_pts
path_pts = []
if (len(table_pts) == 4):
for p1 in table_pts:
for p2 in table_pts:
if (p1 == p2):
break
# Register 2 middle_pts of 4 table legs
if (0.73 < self.getDistance(p1, p2) < 0.83):
path_pts.append(self.getMiddlePoint(p1, p2))
break
print "len(path_pts)", len(path_pts)
print path_pts
# Turn path_pts into map frame
for i in range(len(path_pts)):
path_pts[i] = tf2_geometry_msgs.do_transform_pose(
self.toPoseStamped(path_pts[i], Quaternion(0, 0, 0, 1)),
trans).pose.position
path_pts[i] = Point(path_pts[i].x, path_pts[i].y, 0)
# Record the starting point and heading once
if self.record_once:
self.start_pt = trans.transform.translation
self.record_once = False
# Create a list with distance information between initial_pt and path_pts
distance_list = [
self.getDistance(p, self.start_pt) for p in path_pts
]
# Rearrange the path_pts to be in descending order
path_pts = self.getAscend(path_pts, distance_list)
# Duplicate the path_pts to prevent from being edited
table_pathpts = path_pts[:]
print "table_pathpts: ", len(table_pathpts)
if (len(table_pathpts) == 2):
# Get gradient of the Line of Best Fit
m1 = self.getGradientLineOfBestFit(table_pathpts)
# Insert another point on the line with d[m] away from the 2nd table_pathpts
path_pts.append(
self.getPointOnLine(m1, 1.0, table_pathpts[1],
self.start_pt))
# Visualize the path_pts
self.vizPoint(0, path_pts[0], ColorRGBA(1, 0, 0, 1),
"map") # Red
self.vizPoint(1, path_pts[1], ColorRGBA(1, 1, 0, 1),
"map") # Yellow
self.vizPoint(2, path_pts[2], ColorRGBA(1, 0, 1, 1),
"map") # Magenta
print len(path_pts)
# Find the heading of the table (last 2 path_pts)
heading_q = self.getOrientation(table_pathpts[0],
table_pathpts[1])
# Publish the path
self.route_pub.publish(self.getPath2Table(path_pts, heading_q))
print "-----------------"
def do_euclidian_clustering(self, cloud):
# Euclidean Clustering
white_cloud = pcl_helper.XYZRGB_to_XYZ(cloud)
tree = white_cloud.make_kdtree()
ec = white_cloud.make_EuclideanClusterExtraction()
ec.set_ClusterTolerance(0.3) #0.14, 0.08
ec.set_MinClusterSize(1)
ec.set_MaxClusterSize(3500)
ec.set_SearchMethod(tree)
cluster_indices = ec.Extract()
cluster_color = pcl_helper.random_color_gen()
#cluster_color = pcl_helper.get_color_list(len(cluster_indices))
#print("No. of cones visible at this moment ", len(cluster_indices))
#print(cluster_indices)
cluster_coords = Track()
#cone = TrackCone()
#cluster_coords.cones = []
color_cluster_point_list = []
for j, indices in enumerate(cluster_indices):
x, y, z = 0, 0, 0
cone_clusters = []
for i, indice in enumerate(indices):
color_cluster_point_list.append([white_cloud[indice][0], white_cloud[indice][1], white_cloud[indice][2], pcl_helper.rgb_to_float(cluster_color)])
x = x + white_cloud[indice][0]
y = y + white_cloud[indice][1]
z = z + white_cloud[indice][2]
cone_clusters.append(x/len(indices) + self.carPosX)
cone_clusters.append(y/len(indices) + self.carPosY)
print(cone_clusters)
#print(color_cluster_point_list)
#cone.x = x/len(indices)
#cone.y = y/len(indices)
#cone.type = f'cone{j+1}'
#cluster_coords.cones = cone_clusters
#cluster_coords.data.append(TrackCone(data = cone_clusters))
cluster_coords.data.append(TrackCone(x = x/len(indices) + self.carPosX,
y = y/len(indices) + self.carPosY))
#cluster_coords.cones[j].x = x/len(indices)
#cluster_coords.cones[j].y = y/len(indices)
#cluster_coords.cones[j].type = 'cone{k}'
#print(len(cluster_coords.cones))
#print(len(cluster_coords.cones[0]))
#print("No. of cones visible at this moment ", len(color_cluster_point_list))
cluster_cloud = pcl.PointCloud_PointXYZRGB()
cluster_cloud.from_list(color_cluster_point_list)
ros_cluster_cloud = pcl_helper.pcl_to_ros(cluster_cloud)
return cluster_cloud, cluster_coords
def pc2CB(self, msg):
if self.init:
# Convert ROS PointCloud2 msg to PCL data
cloud = pcl_helper.ros_to_pcl(msg)
# Get transform of "base_link" relative to "map"
trans = self.getTransform("map", "base_link")
# Filter the Cloud
fil_cloud = filtering_helper.do_passthrough_filter(
cloud, 'x', -3.0, 3.0)
fil_cloud = filtering_helper.do_passthrough_filter(
fil_cloud, 'y', -3.0, 3.0)
# Turn cloud into colorless cloud
colorless_cloud = pcl_helper.XYZRGB_to_XYZ(fil_cloud)
# source_cloud = colorless_cloud
# Get groups of cluster of points
clusters = self.getClusters(colorless_cloud, 0.05, 100, 450)
# print "len(clusters)", len(clusters)
source_cloud = self.getCloudFromCluster(clusters[0],
colorless_cloud)
middle_pt = self.getMeanPoint(clusters[0], colorless_cloud)
self.vizPoint(0, middle_pt, ColorRGBA(1, 1, 1, 1), "base_link")
icp = source_cloud.make_IterativeClosestPoint()
converged, transform, estimate, fitness = icp.icp(
source_cloud, self.target_cloud, 100)
# gicp = colorless_cloud.make_GeneralizedIterativeClosestPoint()
# converged, transform, estimate, fitness = gicp.gicp(colorless_cloud, self.target_cloud, 100)
# icp_nl = colorless_cloud.make_IterativeClosestPointNonLinear()
# converged, transform, estimate, fitness = icp_nl.icp_nl(self.target_cloud, colorless_cloud, 100)
# if not converged:
print "converged", converged
print "transform", transform
print "estimate", estimate
print "fitness", fitness
# new_transform = transform.inverse()
# print new_transform
translation = Vector3(transform[0, 3], transform[1, 3],
transform[2, 3])
q = tf.transformations.quaternion_from_matrix(transform)
rotation = Quaternion(q[0], q[1], q[2], q[3])
roll, pitch, yaw = tf.transformations.euler_from_matrix(transform)
print np.rad2deg(yaw)
# print "translation\n", translation
# print "rotation\n", rotation
# transformation = self.toTransformStamped(translation, rotation, "map", "base_link")
print "------------------------------------------------"
# Visualize the cloud
outmsg = pcl_helper.pcl_to_ros(estimate, "map")
outmsg.header = msg.header
self.pc2_pub.publish(outmsg)
def Process_rawPCL(pclpcRawIn):
DebugDumpMsgPCL(pclpcRawIn)
pclRecs = [
] # For dev/debug display. Container for point cloud records: tuple (pclObj, pclName# )
pclRecs.append((pclpcRawIn, "pclpcRawIn"))
pclRecsDownSampled = pclproc.PCLProc_DownSampleVoxels(pclpcRawIn)
pclRecs += pclRecsDownSampled
pclpcDownSampled, pclpcDownSampledName = pclRecsDownSampled[0]
# Region of Interest: PassThrough Filter
recsPassThruZ = pclproc.PCLProc_PassThrough(pclpcDownSampled, 'z', 0.6,
1.1)
pclRecs += recsPassThruZ
pclpcPassZ = recsPassThruZ[0][0]
recsPassThruY = pclproc.PCLProc_PassThrough(pclpcPassZ, 'y', -0.5, 0.5)
pclRecs += recsPassThruY
pclpcPassY = recsPassThruY[0][0]
# RANSAC Table/Object separation
pclRecsRansac = pclproc.PCLProc_Ransac(pclpcPassY)
pclRecs += pclRecsRansac
pclpcPassZIn, pclpcPassZOut = pclRecsRansac[0][0], pclRecsRansac[1][0]
pclpcTable, pclpcObjects = pclpcPassZIn, pclpcPassZOut # Rename for clarity
# Noise reduction
pclpRecsNoNoise = pclproc.PCLProc_Noise(pclpcObjects)
pclRecs += pclpRecsNoNoise
pclpNoColorNoNoise = pclpRecsNoNoise[0][0]
Debug_Publish(pclpNoColorNoNoise)
# Euclidean Clustering
pclpObjectsNoColor = pcl_helper.XYZRGB_to_XYZ(pclpcObjects)
clusterIndices, pclpClusters = pclproc.PCLProc_ExtractClusters(
pclpObjectsNoColor)
#clusterIndices, pclpClusters = pclproc.PCLProc_ExtractClusters(pclpNoColorNoNoise)
labelRecs = []
for index, pts_list in enumerate(clusterIndices):
# Get points for a single object in the overall cluster
pcl_cluster = pclpcObjects.extract(pts_list)
msgPCL_cluster = pcl_helper.pcl_to_ros(
pcl_cluster) # Needed for histograms... would refactor
# Extract histogram features
chists = pclproc.compute_color_histograms(msgPCL_cluster,
g_numHistBinsHSV,
doConvertToHSV=True)
normals = get_normals(msgPCL_cluster)
nhists = pclproc.compute_normal_histograms(normals,
g_numHistBinsNormals)
feature = np.concatenate((chists, nhists))
# CLASSIFY, retrieve the label for the result
# and add it to detected_objects_labels list
prediction = g_clf.predict(g_scaler.transform(feature.reshape(1, -1)))
label = g_encoder.inverse_transform(prediction)[0]
# Accumulate label records for publishing (and labeling detected objects)
label_pos = list(pclpcObjects[pts_list[0]])
label_pos[0] -= 0.1
label_pos[2] += 0.2
labelRecs.append(
(label, label_pos, index,
pcl_cluster)) # Don't like passing pcl_cluster in this Records...
return labelRecs, pclpcObjects, pclpcTable, pclpClusters
def pc2CB(self, msg):
if self.init:
# Convert ROS PointCloud2 msg to PCL data
cloud = pcl_helper.ros_to_pcl(msg)
# Filter the Cloud
fil_cloud = filtering_helper.do_passthrough_filter(cloud, 'x', -2.5, 2.5)
filtered_cloud = filtering_helper.do_passthrough_filter(fil_cloud, 'y', -2.5, 2.5)
# Turn cloud into colorless cloud
colorless_cloud = pcl_helper.XYZRGB_to_XYZ(filtered_cloud)
# Get groups of cluster of points
clusters = self.getClusters(colorless_cloud, 0.05, 10, 1000)
print "len(clusters)", len(clusters)
# Extract charger wall cluster - use for compute Line of Best Fit
if(len(clusters) != 0):
size_clusters = []
for cluster in clusters:
size_clusters.append(len(cluster))
# Find the biggest cluster
max_id = size_clusters.index(max(size_clusters))
wall_cluster = clusters.pop(max_id)
print "len(wall_cluster)", len(wall_cluster)
feature_clusters = clusters
print "len(feature_clusters)", len(feature_clusters)
# Get transform of "base_link" relative to "map"
trans = self.getTransform("map", "base_link")
# Record the starting point and heading once
if self.record_once:
self.start_pt = trans.transform.translation
self.record_once = False
# Get Points of wall cluster
wall_pts = self.getPointsFromCluster(wall_cluster, filtered_cloud)
# Turn wall_pts into map frame
for i in range(len(wall_cluster)):
wall_pts[i] = tf2_geometry_msgs.do_transform_pose(self.toPoseStamped(wall_pts[i], Quaternion(0,0,0,1)), trans).pose.position
# Get gradient of the Line of Best Fit
m1 = self.getGradientLineOfBestFit(wall_pts)
print m1
# Get mean points from feature clusters
mean_pts = []
for cluster in feature_clusters:
mean_pts.append(self.getMeanPointFromCluster(cluster, filtered_cloud))
print "len(mean_pts)", len(mean_pts)
for i in range(len(mean_pts)):
self.vizPoint(i, mean_pts[i], ColorRGBA(1,1,1,1), "base_link")
# Finding 1 middle point from 2 charger feature columns and use it as path_pts
if(len(mean_pts)>=2):
for p1 in mean_pts:
for p2 in mean_pts:
if(p1==p2):
break
# Only register middle_pt of 2 charger feature points
if(0.5 < self.getDistance(p1, p2) < 0.65):
middle_pt = self.getMiddlePoint(p1, p2)
# Turn middle_pt into map frame
middle_pt = tf2_geometry_msgs.do_transform_pose(self.toPoseStamped(middle_pt, Quaternion(0,0,0,1)), trans).pose.position
# Register the middle_pt as one of the path_pts
path_pts = []
path_pts.append(middle_pt)
# Insert another point on the normal line with d[m] away from middle_pt
path_pts.insert(0, self.getPointOnNormalLine(m1, 1.0, middle_pt, self.start_pt))
# Duplicate the path points to prevent from being edited
feature_pathpts = path_pts[:]
# Insert another point on the normal line with d[m] away from middle_pt
path_pts.insert(1, self.getPointOnNormalLine(m1, 0.3, middle_pt, self.start_pt))
# Create a list with distance information between init_pt and path_pts
distance_list = [self.getDistance(p, self.start_pt) for p in path_pts]
# Rearrange the path_pts to be in ascending order
path_pts = self.getAscend(path_pts, distance_list)
print "len(path_pts)", len(path_pts)
# Add the AGV start_pt to the begin of path_pts
# path_pts.insert(0, self.start_pt)
# Remove last point of path_pts (middle pt of 2 charger feature pts)
path_pts = path_pts[:len(path_pts)-1]
# Visualize the path_pts
self.vizPoint(0, path_pts[0], ColorRGBA(1,0,0,1), "map") # Red
self.vizPoint(1, path_pts[1], ColorRGBA(1,1,0,1), "map") # Yellow
# self.vizPoint(2, path_pts[2], ColorRGBA(0,0,1,1), "map") # Blue
# Find the heading of the charger (last 2 path_pts)
heading_q = self.getOrientation(feature_pathpts[-2], feature_pathpts[-1])
# Publish the path
self.route_pub.publish(self.getPath2Charger(path_pts, heading_q))
print "-----------------"