Cognitoware.Robotics.dll
Class KalmanActionModel<U, X>
U: The type of the action being performed on the system.
X: The type of the system x.
System.Object
Cognitoware.Robotics.StateEstimation.KalmanActionModel<U, X>
Implemented Interfaces
Summary
An action model based on the linear system Xt+1 = A*Xt + B*actiont + C + error.
Constructor Summary
Method Summary
Creates a new GaussianMoment using GetMean and GetError.
Equals(Object)
Inherited from System.Object
Finalize()
Inherited from System.Object
Creates an covariance matrix that describes the Gaussian error around the final x mean.
GetHashCode()
Inherited from System.Object
Creates the expected x resulting from performing an action at another x.
GetType()
Inherited from System.Object
MemberwiseClone()
Inherited from System.Object
ToString()
Inherited from System.Object
Details
A linear sensor model is an implementation of GaussianActionModel and RandomConditional.
The expected next x is found using the linear system Xt+1 = A*Xt + B*actiont + C + error.
The error is the constant covariance matrix error.
Because of the requirement to interact with Matrix, U and X must both be of type Vector or inherited from Vector.
Linear action models are also referred to as "Linear Gaussians".
Constructor Details
public KalmanActionModel()
Method Details
public final virtual RandomDistribution<X> ConditionBy(U action, X state)
Creates a new GaussianMoment using GetMean and GetError.
Parameters:
action - The action performed at the start x.
x - The start x.
Returns:
A new Gaussian distribution.
public final virtual Matrix GetActionJacobian(U action, X state)
public final virtual Matrix GetError(U action, X state)
Creates an covariance matrix that describes the Gaussian error around the final x mean.
This function is used to calcualte the covariance of the GaussianMoment returned by ActionModel.ConditionBy.
This function returns the constant value R.
Parameters:
action - The action being performed.
x - The x where the action is performed.
Returns:
A covariance matix that describes the error of the mean.
public final virtual X GetMean(U action, X state)
Returns the expected x resulting from performing an action at another x.
This function is used to calcualte the mean of the GaussianMoment returned by ActionModel.ConditionBy.
Xt+1 = A * Xt + B * U + C.
Parameters:
action - The action performed at the start x.
x - The start x.
public final virtual Matrix GetStateJacobian(U action, X state)