EKFSystemModel

classdef EKFSystemModel < BaseDiscreteSystemModel
%EKFSYSTEMMODEL Class modeling a discrete time system with EKF predictions
%   sysModel = EKFSystemModel(f, Q, J_x) creates a class
%   representing the discrete time system x(t) which evolves via
%      x(t+1) = f(x(t), t) + v
%   where v is gaussian noise with mean 0 and covariance Q
%
%   sysModel = EKFSystemModel(f, Q, J_x, J_v), models the nonlinear system:
%      x(t+1) = f(x(t), t, v)
%
%   sysModel = EKFSystemModel(f, Q, J_x, 'Name1', Value1, 'Name2', Value2, ...)
%   sysModel = EKFSystemModel(f, Q, J_x, J_v, 'Name1', Value1, 'Name2', Value2, ...)
%   
%   Optional Inputs:
%      'Name1', Value1, 'Name2', Value2, ... is a variable length
%      list of parameter/value pairs.
%
%   properties
%      f - function
%      Q - noise covariance matrix
%      J_x - jacobian of f with respect to x
%      J_v - jacobian of f with respect to v
%      isLinearNoise - true if the system model has additive noise
%      
    
    properties
        % inherits f, Q, isLinearNoise from BaseDiscreteSystemModel
        %J_x;  % Jacobian of f with respect to x
        %J_v;  % Jacobian of f with respect to v
    end

    methods
        function sysModel = EKFSystemModel(f, Q, J_x, varargin)
            if nargin == 0
                superargs = {};
            elseif nargin == 1
                superargs = {f};
            elseif nargin == 2
                error('Must specify Jacobian');
            elseif nargin == 3
                superargs = {f, Q, 'J_x', J_x};
            else
                % the fourth argument can be either J_v or start of
                % property/value pairs.
                if isa(varargin{1}, 'function_handle')
                    superargs = [{f, Q, 'J_x', J_x, 'J_v', varargin{1}}, ...
                                  varargin{2:end}];
                else
                    superargs = [{f, Q, 'J_x', J_x}, varargin{:}]
                end
            end
            sysModel = sysModel@BaseDiscreteSystemModel(superargs{:});
        end
        
        function [x, P] = predict(this, x, P, dt, curTime)
            [x, P] = ekfTransform(this.f, this.J_x, x, P, this.Q, ...
                                  curTime, this.isLinearNoise, this.J_v);
        end
    end
    
end