Object used to handle communication between other ROS nodes and CBA. More...

#include <cba.h>

Public Member Functions
	cba_learner ()
	Creates a cba_learner using optional ROS parameters.
void	step ()
	Iterates once trough the CBA algorithm.
virtual	~cba_learner ()
	cba_learner destructor.
Private Member Functions
bool	a_complete_callback (lfd_common::action_complete::Request &req, lfd_common::action_complete::Response &resp)
	a_complete service callback function
prediction *	classify_state ()
	Used to classify the current state.
double	conf_thresh (int l, int db)
	Retrieve the confidence threshold for the given label and decision boundary pair.
double	nearest_neighbor ()
	Used to calculate the nearest neighbor to the current state.
void	state_listener_callback (const lfd_common::state::ConstPtr &msg)
	state_listener topic callback function
void	update_thresholds ()
	Update the distance and confidence thresholds.
Private Attributes
int	a
ros::ServiceServer	a_complete
bool	action_complete
ros::Publisher	add_point
ANNpointArray	ann_data
bool	autonomous_action
ros::Publisher	change_point
ros::ServiceClient	classify
std::vector< conf * >	conf_thresholds
ros::ServiceClient	correction
ros::ServiceClient	demonstration
double	dist_mult
double	dist_thresh
ros::Publisher	execute
int32_t *	labels
int	max_pts
ros::NodeHandle	node
int	pts
float *	s
int	s_size
float *	sc
ros::Subscriber	state_listener

Detailed Description

Object used to handle communication between other ROS nodes and CBA.

The cba_learner handles communication between the agent, human, and CBA via ROS. ROS nodes and services are created and maintained within this object.

Definition at line 75 of file cba.h.

Constructor & Destructor Documentation

cba_learner::cba_learner ( )

Creates a cba_learner using optional ROS parameters.

Creates a cba_learner object that will allocate most of the necessary resources needed and initialize all ROS topics/services. Optional parameters can be supplied via ROS parameters to set the data set size and distance threshold multiplier.

Definition at line 25 of file cba.cpp.

cba_learner::~cba_learner ( ) [virtual]

cba_learner destructor.

Frees the resources used by the CBA learner.

Definition at line 56 of file cba.cpp.

Member Function Documentation

bool cba_learner::a_complete_callback	(	lfd_common::action_complete::Request &	req,
		lfd_common::action_complete::Response &	resp
	)		`[private]`

a_complete service callback function

Set the action complete flag to true and autonomous action flag to false. Furthermore, if the action completed was autonomous and a correction was given by the user, the model is updated appropriately.

Parameters:

req	the request for the a_complete service
resp	the response for the a_complete service; this does not contain any information for this service

Returns:: returns true once completed

Definition at line 364 of file cba.cpp.

prediction * cba_learner::classify_state ( ) [private]

Used to classify the current state.

Query the classifier for the label of the current state. If the communication to the classifier cannot be made, a label of -1 is given with negative infinity confidence.

Returns:: the prediction information for the current state

Definition at line 190 of file cba.cpp.

double cba_learner::conf_thresh	(	int	l,
		int	db
	)		`[private]`

Retrieve the confidence threshold for the given label and decision boundary pair.

Find the confidence threshold for the given label and decision boundary pair. If no threshold value has been set for this pair yet, infinity is returned (the initial threshold).

Parameters:

l	the label value
db	the decision boundary label value

Returns:: the confidence threshold for the given label and decision boundary pair

Definition at line 255 of file cba.cpp.

double cba_learner::nearest_neighbor ( ) [private]

Used to calculate the nearest neighbor to the current state.

Calculate and return the nearest neighbor distance to the current state using ANN. If no data points exist in the data set, infinity is returned.

Returns:: the nearest neighbor distance to the current state

Definition at line 223 of file cba.cpp.

void cba_learner::state_listener_callback ( const lfd_common::state::ConstPtr & msg ) [private]

state_listener topic callback function

Copies the given state vector into a global buffer. If this is the first call to this function, several buffer resources are allocated based on the size of the state vector.

Parameters: