BB2004InvKinWalkingEngine Class Reference

This class implements a walking engine that optimizes its parameters using evolutionary algorithms. More...

#include <BB2004InvKinWalkingEngine.h>

Inheritance diagram for BB2004InvKinWalkingEngine:

[legend]Collaboration diagram for BB2004InvKinWalkingEngine:

[legend]List of all members.


Public Member Functions
	BB2004InvKinWalkingEngine (InvKinWalkingEngine *pEngine)
	Constructor.
	~BB2004InvKinWalkingEngine ()
	Destructor.
virtual bool	executeParameterized (JointData &jointData, const WalkRequest &walkRequest, double positionInWalkingCycle)
	The function executes the walking engine.
bool	handleMessage (InMessage &message)
	The function handles a message sent by the PC.
Private Types
enum	{ numOfIndividuals = 15, numOfFramesPerIndividual = 625, ignoreFrames = 250, useFrames = numOfFramesPerIndividual - ignoreFrames }
Private Member Functions
void	learn (const WalkRequest &walkRequest)
	The function optimizes the gait for the current motion request.
void	setParameters (const BBIndividual &p)
	The function sets a new parameter set using the individual specified.
void	loadParameters (InvKinWalkingParameters &parameters, const char *name)
	The function loads a set of walking parameters.
double	stdV (SensorData::sensors s) const
	The function calculates the standard deviation of the last "useFrames" measurements of a certain sensor.
void	updateOdometry ()
	The function updates the odometry.
Private Attributes
InvKinWalkingEngine *	pEngine
	The address of the actual walking engine.
ERS7EvolveWalkingParameters	parameters [2]
	Two parameter sets that are interchangingly used.
ERS7EvolveWalkingParameters	bestParameters
	The best parameters found so far.
ERS7EvolveWalkingParameters	forward
	Parameters for forward walking.
ERS7EvolveWalkingParameters	fastForward
	Parameters for fast forward walking.
ERS7EvolveWalkingParameters	backward
	Parameters for backward walking.
ERS7EvolveWalkingParameters	stand
	Parameters for standing.
int	currentParameters
	The index that defines which of the two "parameters" are currently used.
int	currentIndividual
	The index of the current individual in the population.
int	currentFrame
	The index of the current frame that measures the performance of the current individual.
int	evolutions
	The number of evolutions done so far.
double	bestFitness
	The best overall fitness found so far.
Pose2D	odometrySpeed
	The current speed of the robot.
double	odometryScaleX
	The scale factor for odometry in x direction.
double	odometryScaleY
	The scale factor for odometry in y direction.
double	odometryScaleRotation
	The scale factor for odometry rotation.
double	odometryCounterRotation
	A factor that compensates rotation resulting from sideward motion.
RingBuffer< Vector2< double >, 100 >	odometrySums [2]
	The current state of the odometers.
FootPositions	prevPositions
	The previous positions of the feet.
RingBuffer< SensorData, numOfFramesPerIndividual >	buffer
	A buffer for previous sensor readings.
Pose2D	measured
	Accumulated odometry for current individual.
Pose2D	target
	Accumulated request motion for current individual.
BBPopulation< BBInvKinIndividual > *	population
	The population of individuals.

Detailed Description

This class implements a walking engine that optimizes its parameters using evolutionary algorithms.

Definition at line 115 of file BB2004InvKinWalkingEngine.h.

Member Enumeration Documentation

anonymous enum [private]

Enumeration values:

numOfIndividuals The number of individuals in the population.

numOfFramesPerIndividual The number of frames tested per individual.

ignoreFrames The number of frames to ignore at the beginning of an individual.

useFrames The number of frames that are assessed per individual.

Definition at line 154 of file BB2004InvKinWalkingEngine.h.