Invert3ChannelController Class Template Reference

class for robot controller that use naglaa's direct matrix inversion for n channels (simple one layer networks) More...

#include <invert3channelcontroller.h>

Inherits InvertController.

Inheritance diagram for Invert3ChannelController:

[legend]Collaboration diagram for Invert3ChannelController:

[legend]List of all members.

Public Member Functions
virtual double	g (double z)
	neuron transfer function
virtual double	g_s (double z)
virtual double	squash (double z)
	squashing function, to protect against to large weight updates
	Invert3ChannelController ()
virtual void	step (const sensor *x_, int sensornumber, motor *y_, int motornumber)
	performs one step (includes learning). Calulates motor commands from sensor inputs.
virtual void	stepNoLearning (const sensor *x_, int sensornumber, motor *y_, int motornumber)
	performs one step without learning. Calulates motor commands from sensor inputs.
Public Attributes
double	A [NUMBER_CHANNELS][NUMBER_CHANNELS]
	model matrix
double	C [NUMBER_CHANNELS][NUMBER_CHANNELS]
	controller matrix
double	h [NUMBER_CHANNELS]
	bias vector
double	x_buffer [BUFFER_SIZE][NUMBER_CHANNELS]
	buffer for input values, x[tbuffersize]=actual value, x[(t-1+buffersize)buffersize]=x(t-1)
double	y_buffer [BUFFER_SIZE][NUMBER_CHANNELS]
	buffer for output values, y[tbuffersize]=actual value(if already calculated!), y[(t-1+buffersize)buffersize]=y(t-1)
double	xsi4E [NUMBER_CHANNELS]
double	xsi4Model [NUMBER_CHANNELS]
int	t
	number of steps, needed for ringbuffer x_buffer
Protected Member Functions
virtual void	inverseMatrix (double Q[NUMBER_CHANNELS][NUMBER_CHANNELS], double Q_1[NUMBER_CHANNELS][NUMBER_CHANNELS])
virtual double	calculateE (const double *x_, const double *x_delay, const double *y_delay)
virtual void	learn (const double *x_, const double *x_delay, const double *y_delay)
virtual void	learnModel (const double *x_actual, double *y_effective)
virtual void	calculateDelayedValues (double source[BUFFER_SIZE][NUMBER_CHANNELS], paramval number_steps_of_delay_, double *target)
	calculate delayed values
virtual void	calculateSmoothValues (double source[BUFFER_SIZE][NUMBER_CHANNELS], paramval number_steps_for_averaging_, double *target)
virtual void	calculateControllerValues (double *x_smooth, double *y)
	calculate controller ouptus
virtual void	putInBuffer (double buffer[BUFFER_SIZE][NUMBER_CHANNELS], const double *values)
virtual int	getInternalParamNames (paramkey *&keylist)
virtual int	getInternalParams (paramval *vallist, int length)

---

Detailed Description

template<int NUMBER_CHANNELS, int BUFFER_SIZE = 2>
class Invert3ChannelController< NUMBER_CHANNELS, BUFFER_SIZE >

class for robot controller that use naglaa's direct matrix inversion for n channels (simple one layer networks)

Implements standart parameters: eps, rho, mu, stepnumber4avg, stepnumber4delay

---

Constructor & Destructor Documentation

Invert3ChannelController

(

)

[inline]

---

Member Function Documentation

virtual void calculateControllerValues	(	double *	x_smooth,
		double *	y
	)			[inline, protected, virtual]

calculate controller ouptus

virtual void calculateDelayedValues	(	double	source[BUFFER_SIZE][NUMBER_CHANNELS],
		paramval	number_steps_of_delay_,
		double *	target
	)			[inline, protected, virtual]

calculate delayed values

virtual double calculateE	(	const double *	x_,
		const double *	x_delay,
		const double *	y_delay
	)			[inline, protected, virtual]

virtual void calculateSmoothValues	(	double	source[BUFFER_SIZE][NUMBER_CHANNELS],
		paramval	number_steps_for_averaging_,
		double *	target
	)			[inline, protected, virtual]

virtual double g

(

double

z

)

[inline, virtual]

neuron transfer function

virtual double g_s

(

double

z

)

[inline, virtual]

virtual int getInternalParamNames

(

paramkey *&

keylist

)

[inline, protected, virtual]

virtual int getInternalParams	(	paramval *	vallist,
		int	length
	)			[inline, protected, virtual]

virtual void inverseMatrix	(	double	Q[NUMBER_CHANNELS][NUMBER_CHANNELS],
		double	Q_1[NUMBER_CHANNELS][NUMBER_CHANNELS]
	)			[inline, protected, virtual]

virtual void learn	(	const double *	x_,
		const double *	x_delay,
		const double *	y_delay
	)			[inline, protected, virtual]

virtual void learnModel	(	const double *	x_actual,
		double *	y_effective
	)			[inline, protected, virtual]

virtual void putInBuffer	(	double	buffer[BUFFER_SIZE][NUMBER_CHANNELS],
		const double *	values
	)			[inline, protected, virtual]

virtual double squash

(

double

z

)

[inline, virtual]

squashing function, to protect against to large weight updates

virtual void step	(	const sensor *	x_,
		int	sensornumber,
		motor *	y_,
		int	motornumber
	)			[inline, virtual]

performs one step (includes learning). Calulates motor commands from sensor inputs.

Reimplemented from AbstractController.

virtual void stepNoLearning	(	const sensor *	x_,
		int	sensornumber,
		motor *	y_,
		int	motornumber
	)			[inline, virtual]

performs one step without learning. Calulates motor commands from sensor inputs.

Reimplemented from AbstractController.

---

Member Data Documentation

double A[NUMBER_CHANNELS][NUMBER_CHANNELS]

model matrix

double C[NUMBER_CHANNELS][NUMBER_CHANNELS]

controller matrix

double h[NUMBER_CHANNELS]

bias vector

int t

number of steps, needed for ringbuffer x_buffer

double x_buffer[BUFFER_SIZE][NUMBER_CHANNELS]

buffer for input values, x[tbuffersize]=actual value, x[(t-1+buffersize)buffersize]=x(t-1)

double xsi4E[NUMBER_CHANNELS]

double xsi4Model[NUMBER_CHANNELS]

double y_buffer[BUFFER_SIZE][NUMBER_CHANNELS]

buffer for output values, y[tbuffersize]=actual value(if already calculated!), y[(t-1+buffersize)buffersize]=y(t-1)

---

The documentation for this class was generated from the following file:

• invert3channelcontroller.h

---