The metric trainer node object. More...

#include <MetricTrainer.h>

Public Member Functions
	MetricTrainer ()
	Creates a new MetricTrainer.
bool	okay () const
	A check for a valid MetricTrainer.
virtual	~MetricTrainer ()
	Cleans up a MetricTrainer.
Private Member Functions
void	trainMetricsCallback (const rail_pick_and_place_msgs::TrainMetricsGoalConstPtr &goal)
	The train metrics action server callback.
Private Attributes
ros::Publisher	aligned_pc_pub_
actionlib::SimpleActionServer < rail_pick_and_place_msgs::TrainMetricsAction >	as_
ros::Publisher	base_pc_pub_
actionlib::SimpleActionClient < rail_pick_and_place_msgs::GetYesNoFeedbackAction >	get_yes_and_no_feedback_ac_
graspdb::Client *	graspdb_
ros::NodeHandle	node_
bool	okay_
ros::NodeHandle	private_node_

Detailed Description

The metric trainer node object.

The metric trainer allows for generating data sets for training registration metric decision trees. An action server is used to provide the object name and files are dumped to "registration_metrics.txt".

Definition at line 36 of file MetricTrainer.h.

Constructor & Destructor Documentation

MetricTrainer::MetricTrainer ( )

Creates a new MetricTrainer.

Creates a new MetricTrainer with the associated topics and action servers.

Definition at line 23 of file MetricTrainer.cpp.

MetricTrainer::~MetricTrainer ( ) [virtual]

Cleans up a MetricTrainer.

Cleans up any connections used by the MetricTrainer.

Definition at line 57 of file MetricTrainer.cpp.

Member Function Documentation

bool MetricTrainer::okay ( ) const

A check for a valid MetricTrainer.

This function will return true if the appropriate connections were created successfully during initialization.

Returns:: True if the appropriate connections were created successfully during initialization.

Definition at line 65 of file MetricTrainer.cpp.

void MetricTrainer::trainMetricsCallback ( const rail_pick_and_place_msgs::TrainMetricsGoalConstPtr & goal ) [private]

The train metrics action server callback.

This function starts by loading any grasps with the object name from the goal. All pairs of grasps are joined and published to two different point cloud topics (~/base_pc and ~/aligned_pc). User feedback is requested via an action client call to ~/get_yes_no_feedback. The metrics and the responses are recoreded in the data file.

Parameters: