High level interface for the localization of grasp hypotheses and handles. More...

#include <localization.h>

Classes
struct	UniqueVectorComparator
	Comparator for checking uniqueness of two 3D-vectors. More...
Public Member Functions
void	createVisualsPub (ros::NodeHandle &node, double marker_lifetime, const std::string &frame)
	Set the publisher for Rviz visualization, the lifetime of visual markers, and the frame associated with the grasps.
std::vector< Handle >	findHandles (const std::vector< GraspHypothesis > &hand_list, int min_inliers, double min_length)
	Find handles given a list of grasp hypotheses.
const Eigen::Matrix4d &	getCameraTransform (bool is_left)
	Return the camera pose of one given camera.
	Localization ()
	Default Constructor.
	Localization (int num_threads, bool filters_boundaries, int plotting_mode)
	Constructor.
std::vector< GraspHypothesis >	localizeHands (const PointCloud::Ptr &cloud_in, int size_left, const std::vector< int > &indices, bool calculates_antipodal, bool uses_clustering)
	Localize hands in a given point cloud.
std::vector< GraspHypothesis >	localizeHands (const std::string &pcd_filename_left, const std::string &pcd_filename_right, bool calculates_antipodal=false, bool uses_clustering=false)
	Localize hands given two point cloud files.
std::vector< GraspHypothesis >	localizeHands (const std::string &pcd_filename_left, const std::string &pcd_filename_right, const std::vector< int > &indices, bool calculates_antipodal=false, bool uses_clustering=false)
	Localize hands given two point cloud files and a set of point cloud indices.
std::vector< GraspHypothesis >	predictAntipodalHands (const std::vector< GraspHypothesis > &hand_list, const std::string &svm_filename)
	Predict antipodal grasps given a list of grasp hypotheses.
void	setCameraTransforms (const Eigen::Matrix4d &cam_tf_left, const Eigen::Matrix4d &cam_tf_right)
	Set the camera poses.
void	setFingerWidth (double finger_width)
	Set the finger width of the robot hand.
void	setHandDepth (double hand_depth)
	Set the hand depth of the robot hand.
void	setHandHeight (double hand_height)
	Set the height of the robot hand.
void	setHandOuterDiameter (double hand_outer_diameter)
	Set the maximum aperture of the robot hand.
void	setInitBite (double init_bite)
	Set the initial "bite" of the robot hand (usually the minimum object "height").
void	setNeighborhoodRadiusHands (double nn_radius_hands)
	Set the radius to be used for the point neighborhood search in the hand search.
void	setNeighborhoodRadiusTaubin (double nn_radius_taubin)
	Set the radius to be used for the point neighborhood search in the quadric fit.
void	setNumSamples (int num_samples)
	Set the number of samples to be used for the search.
void	setWorkspace (const Eigen::VectorXd &workspace)
	Set the dimensions of the robot's workspace.
Static Public Attributes
static const int	NO_PLOTTING = 0
	no plotting
static const int	PCL_PLOTTING = 1
	plotting in PCL
static const int	PCL_PLOTTING_FINGERS = 2
	plotting in PCL, plots hands with fingers
static const int	RVIZ_PLOTTING = 3
	plotting in Rviz
Private Member Functions
std::vector< GraspHypothesis >	filterHands (const std::vector< GraspHypothesis > &hand_list)
	Filter out grasp hypotheses that are close to the workspace boundaries.
void	filterWorkspace (const PointCloud::Ptr &cloud_in, const Eigen::VectorXi &pts_cam_source_in, PointCloud::Ptr &cloud_out, Eigen::VectorXi &pts_cam_source_out)
	Filter out points in the point cloud that lie outside the workspace dimensions and keep track of the camera source for each point that is not filtered out.
Eigen::Vector3i	floorVector (const Eigen::Vector3d &a)
	Round a 3D-vector down to the closest, smaller integers.
void	voxelizeCloud (const PointCloud::Ptr &cloud_in, const Eigen::VectorXi &pts_cam_source_in, PointCloud::Ptr &cloud_out, Eigen::VectorXi &pts_cam_source_out, double cell_size)
	Voxelize the point cloud and keep track of the camera source for each voxel.
Private Attributes
Eigen::Matrix4d	cam_tf_left_
Eigen::Matrix4d	cam_tf_right_
	the camera poses
PointCloud::Ptr	cloud_
	the input point cloud
Eigen::Matrix3Xd	cloud_normals_
	the normals for each point in the point cloud
bool	filters_boundaries_
	whether grasp hypotheses close to the workspace boundaries are filtered out
double	finger_width_
	width of the fingers
double	hand_depth_
	hand depth (finger length)
double	hand_height_
	hand height
double	hand_outer_diameter_
	maximum hand aperture
double	init_bite_
	initial bite
double	nn_radius_hands_
	the radius of the neighborhood search used in the quadric fit
double	nn_radius_taubin_
	the radius of the neighborhood search used in the grasp hypothesis search
int	num_samples_
	the number of samples used in the search
int	num_threads_
	the number of CPU threads used in the search
Plot	plot_
	the plot object
bool	plots_camera_sources_
	whether the camera source is plotted for each point in the point cloud
int	plotting_mode_
	what plotting mode is used
std::string	visuals_frame_
	visualization frame for Rviz
Eigen::VectorXd	workspace_
	the robot's workspace dimensions

Detailed Description

High level interface for the localization of grasp hypotheses and handles.

Localization class

This class provides a high level interface to search for grasp hypotheses, antipodal grasps, and handles. It can also preprocess the point clouds.

Definition at line 68 of file localization.h.

Constructor & Destructor Documentation

Localization::Localization ( ) [inline]

Default Constructor.

Definition at line 75 of file localization.h.

Localization::Localization	(	int	num_threads,
		bool	filters_boundaries,
		int	plotting_mode
	)		`[inline]`

Constructor.

Parameters:

num_threads	the number of threads to be used in the search
filter_boundaries	whether grasp hypotheses that are close to the point cloud boundaries are filtered out
plots_hands	whether grasp hypotheses are plotted

Definition at line 84 of file localization.h.

Member Function Documentation

void Localization::createVisualsPub	(	ros::NodeHandle &	node,
		double	marker_lifetime,
		const std::string &	frame
	)		`[inline]`

Set the publisher for Rviz visualization, the lifetime of visual markers, and the frame associated with the grasps.

Parameters:

node	the ROS node
marker_lifetime	the lifetime of each visual marker
frame	the frame to which the grasps belong

Definition at line 255 of file localization.h.

std::vector< GraspHypothesis > Localization::filterHands ( const std::vector< GraspHypothesis > & hand_list ) [private]

Filter out grasp hypotheses that are close to the workspace boundaries.

Parameters:

hand_list the list of grasp hypotheses to be filtered

Returns:: the list of grasp hypotheses that are not filtered out

Definition at line 364 of file localization.cpp.

void Localization::filterWorkspace	(	const PointCloud::Ptr &	cloud_in,
		const Eigen::VectorXi &	pts_cam_source_in,
		PointCloud::Ptr &	cloud_out,
		Eigen::VectorXi &	pts_cam_source_out
	)		`[private]`

Filter out points in the point cloud that lie outside the workspace dimensions and keep track of the camera source for each point that is not filtered out.

Parameters:

[in]	cloud_in	the point cloud to be filtered
[in]	pts_cam_source_in	the camera source for each point in the point cloud
[out]	cloud_out	the filtered point cloud
[out]	pts_cam_source_out	the camera source for each point in the filtered cloud

Definition at line 216 of file localization.cpp.

std::vector< Handle > Localization::findHandles	(	const std::vector< GraspHypothesis > &	hand_list,
		int	min_inliers,
		double	min_length
	)

Find handles given a list of grasp hypotheses.

Parameters:

hand_list	the list of grasp hypotheses
min_inliers	the minimum number of handle inliers
min_length	the minimum length of the handle
is_plotting	whether the handles are plotted

Definition at line 390 of file localization.cpp.

Eigen::Vector3i Localization::floorVector ( const Eigen::Vector3d & a ) [private]

Round a 3D-vector down to the closest, smaller integers.

Parameters:

a	the 3D-vector to be rounded down

Returns:: the rounded down 3D-vector

Definition at line 357 of file localization.cpp.

const Eigen::Matrix4d& Localization::getCameraTransform ( bool is_left ) [inline]

Return the camera pose of one given camera.

Parameters:

is_left true if the pose of the left camera is wanted, false if the pose of the right camera is wanted

Returns:: the pose of the camera specified by is_left

Definition at line 159 of file localization.h.

std::vector< GraspHypothesis > Localization::localizeHands	(	const PointCloud::Ptr &	cloud_in,
		int	size_left,
		const std::vector< int > &	indices,
		bool	calculates_antipodal,
		bool	uses_clustering
	)

Localize hands in a given point cloud.

Parameters:

cloud_in	the input point cloud
indices	the set of point cloud indices for which point neighborhoods are found
calculates_antipodal	whether the grasp hypotheses are checked for being antipodal
uses_clustering	whether clustering is used for processing the point cloud

Returns:: the list of grasp hypotheses found

Definition at line 3 of file localization.cpp.

std::vector< GraspHypothesis > Localization::localizeHands	(	const std::string &	pcd_filename_left,
		const std::string &	pcd_filename_right,
		bool	calculates_antipodal = `false`,
		bool	uses_clustering = `false`
	)

Localize hands given two point cloud files.

Parameters:

pcd_filename_left	the first point cloud file location and name
pcd_filename_right	the second point cloud file location and name
calculates_antipodal	whether the grasp hypotheses are checked for being antipodal
uses_clustering	whether clustering is used for processing the point cloud

Returns:: the list of grasp hypotheses found

Definition at line 169 of file localization.cpp.

std::vector< GraspHypothesis > Localization::localizeHands	(	const std::string &	pcd_filename_left,
		const std::string &	pcd_filename_right,
		const std::vector< int > &	indices,
		bool	calculates_antipodal = `false`,
		bool	uses_clustering = `false`
	)

Localize hands given two point cloud files and a set of point cloud indices.

Parameters:

pcd_filename_left	the first point cloud file location and name
pcd_filename_right	the second point cloud file location and name
indices	the set of point cloud indices for which point neighborhoods are found
calculates_antipodal	whether the grasp hypotheses are checked for being antipodal
uses_clustering	whether clustering is used for processing the point cloud

Returns:: the list of grasp hypotheses found

Definition at line 176 of file localization.cpp.

std::vector< GraspHypothesis > Localization::predictAntipodalHands	(	const std::vector< GraspHypothesis > &	hand_list,
		const std::string &	svm_filename
	)

Predict antipodal grasps given a list of grasp hypotheses.

Parameters:

hand_list	the list of grasp hypotheses
svm_filename	the location and filename of the SVM

Definition at line 142 of file localization.cpp.

void Localization::setCameraTransforms	(	const Eigen::Matrix4d &	cam_tf_left,
		const Eigen::Matrix4d &	cam_tf_right
	)		`[inline]`

Set the camera poses.

Parameters:

cam_tf_left	the pose of the left camera
cam_tf_right	the pose of the right camera

Definition at line 148 of file localization.h.

void Localization::setFingerWidth ( double finger_width ) [inline]

Set the finger width of the robot hand.

Parameters:

finger_width the finger width

Definition at line 207 of file localization.h.

void Localization::setHandDepth ( double hand_depth ) [inline]

Set the hand depth of the robot hand.

Parameters:

hand_depth the hand depth of the robot hand (usually the finger length)

Definition at line 216 of file localization.h.

void Localization::setHandHeight ( double hand_height ) [inline]

Set the height of the robot hand.

Parameters:

hand_height the height of the robot hand, the hand extends plus/minus this value along the hand axis

Definition at line 243 of file localization.h.

void Localization::setHandOuterDiameter ( double hand_outer_diameter ) [inline]

Set the maximum aperture of the robot hand.

Parameters:

hand_outer_diameter the maximum aperture of the robot hand

Definition at line 225 of file localization.h.

void Localization::setInitBite ( double init_bite ) [inline]

Set the initial "bite" of the robot hand (usually the minimum object "height").

Parameters:

init_bite the initial "bite" of the robot hand (

See also:: FingerHand)

Definition at line 234 of file localization.h.

void Localization::setNeighborhoodRadiusHands ( double nn_radius_hands ) [inline]

Set the radius to be used for the point neighborhood search in the hand search.

Parameters:

nn_radius_hands the radius to be used for the point neighborhood search

Definition at line 189 of file localization.h.

void Localization::setNeighborhoodRadiusTaubin ( double nn_radius_taubin ) [inline]

Set the radius to be used for the point neighborhood search in the quadric fit.

Parameters:

nn_radius_hands the radius to be used for the point neighborhood search

Definition at line 198 of file localization.h.

void Localization::setNumSamples ( int num_samples ) [inline]

Set the number of samples to be used for the search.

Parameters:

num_samples the number of samples to be used for the search

Definition at line 180 of file localization.h.

void Localization::setWorkspace ( const Eigen::VectorXd & workspace ) [inline]

Set the dimensions of the robot's workspace.

Parameters:

workspace 1x6 vector containing the robot's workspace dimensions

Definition at line 171 of file localization.h.

void Localization::voxelizeCloud	(	const PointCloud::Ptr &	cloud_in,
		const Eigen::VectorXi &	pts_cam_source_in,
		PointCloud::Ptr &	cloud_out,
		Eigen::VectorXi &	pts_cam_source_out,
		double	cell_size
	)		`[private]`

Voxelize the point cloud and keep track of the camera source for each voxel.

Parameters:

[in]	cloud_in	the point cloud to be voxelized
[in]	pts_cam_source_in	the camera source for each point in the point cloud
[out]	cloud_out	the voxelized point cloud
[out]	pts_cam_source_out	the camera source for each point in the voxelized cloud
[in]	cell_size	the size of each voxel