hinted_handle_estimator_nodelet.cpp
Go to the documentation of this file.
1 // -*- mode: c++ -*-
2 /*********************************************************************
3  * Software License Agreement (BSD License)
4  *
5  * Copyright (c) 2015, JSK Lab
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  *
12  * * Redistributions of source code must retain the above copyright
13  * notice, this list of conditions and the following disclaimer.
14  * * Redistributions in binary form must reproduce the above
15  * copyright notice, this list of conditions and the following
16  * disclaimer in the documentation and/o2r other materials provided
17  * with the distribution.
18  * * Neither the name of the JSK Lab nor the names of its
19  * contributors may be used to endorse or promote products derived
20  * from this software without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
23  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
24  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
25  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
26  * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
27  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
28  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
29  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
30  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
32  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
33  * POSSIBILITY OF SUCH DAMAGE.
34  *********************************************************************/
35 
36 #define BOOST_PARAMETER_MAX_ARITY 7
39 #include <pcl/features/normal_3d.h>
40 #include <pcl/kdtree/kdtree_flann.h>
41 #include <pcl/common/transforms.h>
42 #include <pcl/filters/passthrough.h>
45 #include <tf/tf.h>
46 #include <math.h>
47 
48 using namespace std;
49 
50 inline float NORM(float vx, float vy, float vz){return sqrt(vx*vx+vy*vy+vz*vz);}
51 
52 namespace jsk_pcl_ros
53 {
54  void HintedHandleEstimator::onInit()
55  {
56  DiagnosticNodelet::onInit();
57  pub_pose_ = advertise<geometry_msgs::PoseStamped>(
58  *pnh_, "handle_pose", 1);
59  pub_length_ = advertise<std_msgs::Float64>(
60  *pnh_, "handle_length", 1);
61  pub_handle_ = advertise<jsk_recognition_msgs::SimpleHandle>(
62  *pnh_, "handle", 1);
63  pub_debug_marker_ = advertise<visualization_msgs::Marker>(*pnh_, "debug_marker", 1);
64  pub_debug_marker_array_ = advertise<visualization_msgs::MarkerArray>(*pnh_, "debug_marker_array", 1);
65  handle = handle_model();
66  pnh_->param("finger_l", handle.finger_l, 0.03);
67  pnh_->param("finger_d", handle.finger_d, 0.02);
68  pnh_->param("finger_w", handle.finger_w, 0.01);
69  pnh_->param("arm_l", handle.arm_l, 0.05);
70  pnh_->param("arm_d", handle.arm_d, 0.02);
71  pnh_->param("arm_w", handle.arm_w, 0.1);
72  onInitPostProcess();
73  }
74 
75  void HintedHandleEstimator::subscribe()
76  {
77  sub_cloud_.subscribe(*pnh_, "cloud", 1);
78  sub_point_.subscribe(*pnh_, "point", 1);
79  sync_ = boost::make_shared<message_filters::Synchronizer<SyncPolicy> >(100);
80  sync_->connectInput(sub_cloud_, sub_point_);
81  sync_->registerCallback(boost::bind(&HintedHandleEstimator::estimate, this, _1, _2));
82  }
83 
84  void HintedHandleEstimator::unsubscribe()
85  {
86  sub_cloud_.unsubscribe();
87  sub_point_.unsubscribe();
88  }
89 
90  void HintedHandleEstimator::estimate(
91  const sensor_msgs::PointCloud2::ConstPtr& cloud_msg,
92  const geometry_msgs::PointStampedConstPtr &point_msg)
93  {
94  boost::mutex::scoped_lock lock(mutex_);
95  pcl::PointCloud<pcl::PointXYZ>::Ptr cloud(new pcl::PointCloud<pcl::PointXYZ>);
96  pcl::PointCloud<pcl::Normal>::Ptr cloud_normals(new pcl::PointCloud<pcl::Normal>);
97  pcl::PassThrough<pcl::PointXYZ> pass;
98  int K = 1;
99  std::vector<int> pointIdxNKNSearch(K);
100  std::vector<float> pointNKNSquaredDistance(K);
101  pcl::search::KdTree<pcl::PointXYZ>::Ptr kd_tree(new pcl::search::KdTree<pcl::PointXYZ>);
102 
103  pcl::fromROSMsg(*cloud_msg, *cloud);
104  geometry_msgs::PointStamped transed_point;
106  try
107  {
108  listener_.waitForTransform(cloud->header.frame_id, point_msg->header.frame_id, now, ros::Duration(1.0));
109  listener_.transformPoint(cloud->header.frame_id, now, *point_msg, point_msg->header.frame_id, transed_point);
110  }
112  {
113  ROS_ERROR("%s", ex.what());
114  return;
115  }
116  pcl::PointXYZ searchPoint;
117  searchPoint.x = transed_point.point.x;
118  searchPoint.y = transed_point.point.y;
119  searchPoint.z = transed_point.point.z;
120 
121  //remove too far cloud
122  pass.setInputCloud(cloud);
123  pass.setFilterFieldName("x");
124  pass.setFilterLimits(searchPoint.x - 3*handle.arm_w, searchPoint.x + 3*handle.arm_w);
125  pass.filter(*cloud);
126  pass.setInputCloud(cloud);
127  pass.setFilterFieldName("y");
128  pass.setFilterLimits(searchPoint.y - 3*handle.arm_w, searchPoint.y + 3*handle.arm_w);
129  pass.filter(*cloud);
130  pass.setInputCloud(cloud);
131  pass.setFilterFieldName("z");
132  pass.setFilterLimits(searchPoint.z - 3*handle.arm_w, searchPoint.z + 3*handle.arm_w);
133  pass.filter(*cloud);
134 
135  if(cloud->points.size() < 10){
136  ROS_INFO("points are too small");
137  return;
138  }
139  if(1){ //estimate_normal
140  pcl::NormalEstimation<pcl::PointXYZ, pcl::Normal> ne;
141  ne.setInputCloud(cloud);
142  ne.setSearchMethod(kd_tree);
143  ne.setRadiusSearch(0.02);
144  ne.setViewPoint(0, 0, 0);
145  ne.compute(*cloud_normals);
146  }
147  else{ //use normal of msg
148 
149  }
150  if(! (kd_tree->nearestKSearch (searchPoint, K, pointIdxNKNSearch, pointNKNSquaredDistance) > 0)){
151  ROS_INFO("kdtree failed");
152  return;
153  }
154  float x = cloud->points[pointIdxNKNSearch[0]].x;
155  float y = cloud->points[pointIdxNKNSearch[0]].y;
156  float z = cloud->points[pointIdxNKNSearch[0]].z;
157  float v_x = cloud_normals->points[pointIdxNKNSearch[0]].normal_x;
158  float v_y = cloud_normals->points[pointIdxNKNSearch[0]].normal_y;
159  float v_z = cloud_normals->points[pointIdxNKNSearch[0]].normal_z;
160  double theta = acos(v_x);
161  // use normal for estimating handle direction
162  tf::Quaternion normal(0, v_z/NORM(0, v_y, v_z) * cos(theta/2), -v_y/NORM(0, v_y, v_z) * cos(theta/2), sin(theta/2));
163  tf::Quaternion final_quaternion = normal;
164  double min_theta_index = 0;
165  double min_width = 100;
166  tf::Quaternion min_qua(0, 0, 0, 1);
167  visualization_msgs::Marker debug_hand_marker;
168  debug_hand_marker.header = cloud_msg->header;
169  debug_hand_marker.ns = string("debug_grasp");
170  debug_hand_marker.id = 0;
171  debug_hand_marker.type = visualization_msgs::Marker::LINE_LIST;
172  debug_hand_marker.pose.orientation.w = 1;
173  debug_hand_marker.scale.x=0.003;
174  tf::Matrix3x3 best_mat;
175  //search 180 degree and calc the shortest direction
176  for(double theta_=0; theta_<3.14/2;
177  theta_+=3.14/2/30){
178  tf::Quaternion rotate_(sin(theta_), 0, 0, cos(theta_));
179  tf::Quaternion temp_qua = normal * rotate_;
180  tf::Matrix3x3 temp_mat(temp_qua);
181  geometry_msgs::Pose pose_respected_to_tf;
182  pose_respected_to_tf.position.x = x;
183  pose_respected_to_tf.position.y = y;
184  pose_respected_to_tf.position.z = z;
185  pose_respected_to_tf.orientation.x = temp_qua.getX();
186  pose_respected_to_tf.orientation.y = temp_qua.getY();
187  pose_respected_to_tf.orientation.z = temp_qua.getZ();
188  pose_respected_to_tf.orientation.w = temp_qua.getW();
189  Eigen::Affine3d box_pose_respected_to_cloud_eigend;
190  tf::poseMsgToEigen(pose_respected_to_tf, box_pose_respected_to_cloud_eigend);
191  Eigen::Affine3d box_pose_respected_to_cloud_eigend_inversed
192  = box_pose_respected_to_cloud_eigend.inverse();
193  Eigen::Matrix4f box_pose_respected_to_cloud_eigen_inversed_matrixf;
194  Eigen::Matrix4d box_pose_respected_to_cloud_eigen_inversed_matrixd
195  = box_pose_respected_to_cloud_eigend_inversed.matrix();
196  jsk_recognition_utils::convertMatrix4<Eigen::Matrix4d, Eigen::Matrix4f>(
197  box_pose_respected_to_cloud_eigen_inversed_matrixd,
198  box_pose_respected_to_cloud_eigen_inversed_matrixf);
199  Eigen::Affine3f offset = Eigen::Affine3f(box_pose_respected_to_cloud_eigen_inversed_matrixf);
200  pcl::PointCloud<pcl::PointXYZ>::Ptr output_cloud(new pcl::PointCloud<pcl::PointXYZ>);
201  pcl::transformPointCloud(*cloud, *output_cloud, offset);
202 
203  pcl::PassThrough<pcl::PointXYZ> pass;
204  pcl::PointCloud<pcl::PointXYZ>::Ptr points_z(new pcl::PointCloud<pcl::PointXYZ>), points_yz(new pcl::PointCloud<pcl::PointXYZ>), points_xyz(new pcl::PointCloud<pcl::PointXYZ>);
205  pass.setInputCloud(output_cloud);
206  pass.setFilterFieldName("y");
207  pass.setFilterLimits(-handle.arm_w*2, handle.arm_w*2);
208  pass.filter(*points_z);
209  pass.setInputCloud(points_z);
210  pass.setFilterFieldName("z");
211  pass.setFilterLimits(-handle.finger_d, handle.finger_d);
212  pass.filter(*points_yz);
213  pass.setInputCloud(points_yz);
214  pass.setFilterFieldName("x");
215  pass.setFilterLimits(-(handle.arm_l-handle.finger_l), handle.finger_l);
216  pass.filter(*points_xyz);
217  pcl::KdTreeFLANN<pcl::PointXYZ> kdtree;
218  for(size_t index=0; index<points_xyz->size(); index++){
219  points_xyz->points[index].x = points_xyz->points[index].z = 0;
220  }
221  if(points_xyz->points.size() == 0){ROS_INFO("points are empty");return;}
222  kdtree.setInputCloud(points_xyz);
223  std::vector<int> pointIdxRadiusSearch;
224  std::vector<float> pointRadiusSquaredDistance;
225  pcl::PointXYZ search_point_tree;
226  search_point_tree.x=search_point_tree.y=search_point_tree.z=0;
227  if( kdtree.radiusSearch(search_point_tree, 10, pointIdxRadiusSearch, pointRadiusSquaredDistance) > 0 ){
228  double before_w=10, temp_w;
229  for(size_t index = 0; index < pointIdxRadiusSearch.size(); ++index){
230  temp_w =sqrt(pointRadiusSquaredDistance[index]);
231  if(temp_w - before_w > handle.finger_w*2){
232  break; // there are small space for finger
233  }
234  before_w=temp_w;
235  }
236  if(before_w < min_width){
237  min_theta_index = theta_;
238  min_width = before_w;
239  min_qua = temp_qua;
240  best_mat = temp_mat;
241  }
242  //for debug view
243  geometry_msgs::Point temp_point;
244  std_msgs::ColorRGBA temp_color;
245  temp_color.r=0; temp_color.g=0; temp_color.b=1; temp_color.a=1;
246  temp_point.x=x-temp_mat.getColumn(1)[0] * before_w;
247  temp_point.y=y-temp_mat.getColumn(1)[1] * before_w;
248  temp_point.z=z-temp_mat.getColumn(1)[2] * before_w;
249  debug_hand_marker.points.push_back(temp_point);
250  debug_hand_marker.colors.push_back(temp_color);
251  temp_point.x+=2*temp_mat.getColumn(1)[0] * before_w;
252  temp_point.y+=2*temp_mat.getColumn(1)[1] * before_w;
253  temp_point.z+=2*temp_mat.getColumn(1)[2] * before_w;
254  debug_hand_marker.points.push_back(temp_point);
255  debug_hand_marker.colors.push_back(temp_color);
256  }
257  }
258  geometry_msgs::PoseStamped handle_pose_stamped;
259  handle_pose_stamped.header = cloud_msg->header;
260  handle_pose_stamped.pose.position.x = x;
261  handle_pose_stamped.pose.position.y = y;
262  handle_pose_stamped.pose.position.z = z;
263  handle_pose_stamped.pose.orientation.x = min_qua.getX();
264  handle_pose_stamped.pose.orientation.y = min_qua.getY();
265  handle_pose_stamped.pose.orientation.z = min_qua.getZ();
266  handle_pose_stamped.pose.orientation.w = min_qua.getW();
267  std_msgs::Float64 min_width_msg;
268  min_width_msg.data = min_width;
269  pub_pose_.publish(handle_pose_stamped);
270  pub_debug_marker_.publish(debug_hand_marker);
271  pub_debug_marker_array_.publish(make_handle_array(handle_pose_stamped, handle));
272  jsk_recognition_msgs::SimpleHandle simple_handle;
273  simple_handle.header = handle_pose_stamped.header;
274  simple_handle.pose = handle_pose_stamped.pose;
275  simple_handle.handle_width = min_width;
276  pub_handle_.publish(simple_handle);
277  }
278 }
279 
282 
283 
void publish(const boost::shared_ptr< M > &message) const
double cos()
double sin()
PLUGINLIB_EXPORT_CLASS(jsk_pcl_ros::HintedHandleEstimator, nodelet::Nodelet)
void poseMsgToEigen(const geometry_msgs::Pose &m, Eigen::Affine3d &e)
void fromROSMsg(const sensor_msgs::PointCloud2 &cloud, pcl::PointCloud< T > &pcl_cloud)
TFSIMD_FORCE_INLINE const tfScalar & getW() const
K
TFSIMD_FORCE_INLINE const tfScalar & x() const
unsigned int index
TFSIMD_FORCE_INLINE const tfScalar & z() const
double sqrt()
#define ROS_INFO(...)
TFSIMD_FORCE_INLINE const tfScalar & y() const
ros::Publisher pub_pose_
visualization_msgs::MarkerArray make_handle_array(geometry_msgs::PoseStamped pose, handle_model handle)
INLINE Rall1d< T, V, S > acos(const Rall1d< T, V, S > &x)
static Time now()
int theta
boost::mutex mutex_
cloud
#define ROS_ERROR(...)
float NORM(float vx, float vy, float vz)
TFSIMD_FORCE_INLINE Vector3 getColumn(int i) const


jsk_pcl_ros
Author(s): Yohei Kakiuchi
autogenerated on Mon May 3 2021 03:03:46