Créez un modèle de modèle de capteur à l'aide de la fonction objet insCreateMotionModel . Spécifiez le nom de la classe comme newSensorModel.
Dans l'éditeur MATLAB , un fichier sans titre s'ouvre avec sa définition de classe. Après avoir modifié la définition de la classe, enregistrez le fichier de classe et utilisez un objet de la classe avec l'objet insEKF .
classdef newSensorModel < positioning.INSMotionModel
%newSensorModel Template for motion model using insEKF
% Customize this motion model and use it with the insEKF to fuse
% data.
%
% Example:
% filt = insEKF(newSensorModel);
%
% See also insEKF, positioning.INSMotionModel.
% Generated on 13-Mar-2022 11:37:21
properties (Constant)
State1Length = 1 % Length of motion model state State1
State2Length = 2 % Length of motion model state State2
end
methods
function s = modelstates(motion, opts)
%modelstates Define the tracked states for this motion model
% MODELSTATES returns a struct which describes the
% states used by this motion model and tracked by the insEKF
% filter object. The field names describe the individual state
% quantities, and you can access the estimates of those
% quantities through the statesparts function. The values of
% the struct determine the size and default values of the
% state vector. The input OPTS is the insOptions object used
% to build the filter.
%
% See also insEKF, positioning.INSMotionModel.
% Preallocate a struct with fields State1 and State2.
% Overwrite the fields with different default values if
% needed.
s = struct("State1", zeros(1, motion.State1Length, opts.Datatype), ...
"State2", zeros(1, motion.State2Length, opts.Datatype));
end
function statesdot = stateTransition(motion, filt, dt, varargin)
%stateTransition State transition for motion states
% STATETRANSITION returns a struct with identical fields
% as the output of the modelstates function. The
% returned struct describes the per-state transition function
% for the motion model states.
%
% This function is called by the insEKF object FILT when the
% PREDICT method of the FILT function is called. The DT and
% varargin inputs are the corresponding inputs to the
% predict method of the insEKF object.
%
% *** THIS METHOD IS OPTIONAL ***
% If you delete this method, the model states will be
% constant. In this case, also delete the
% stateTransitionJacobian method.
%
% See also insEKF, positioning.INSMotionModel.
% Set statesdot.State1 to the derivative of State1 with
% respect to time. If State1 is constant overtime, leave the
% following line unchanged.
statesdot.State1 = zeros(1, motion.State1Length, "like", filt.State);
% Set statesdot.State2 to the derivative of State2 with
% respect to time. If State2 is constant overtime, leave the
% following line unchanged.
statesdot.State2 = zeros(1, motion.State2Length, "like", filt.State);
end
function dfdx = stateTransitionJacobian(motion, filt, dt, varargin)
%stateTransitionJacobian Jacobian of the stateTransition function
% STATETRANSITIONJACOBIAN returns a struct with identical
% fields as modelstates and describes the Jacobian of the
% per-state transition function relative to the State
% property of FILT. Each field value of STATESDOT should be a
% M-by-numel(FILT.State) row vector, representing the partial
% derivatives of that field's state transition function
% relative to the state vector.
%
% This function is called by the insEKF object FILT when the
% PREDICT method of the FILT function is called. The DT and
% varargin inputs are the corresponding inputs to the
% predict method.
%
% *** THIS METHOD IS OPTIONAL ***
% If this method is not implemented, a numerical Jacobian
% will be used instead.
%
% See also insEKF, positioning.INSMotionModel.
N = numel(filt.State);
dfdx.State1 = zeros(motion.State1Length, N, "like", filt.State);
dfdx.State2 = zeros(motion.State2Length, N, "like", filt.State);
% Create indexing
s1idx = stateinfo(filt, "State1");
s2idx = stateinfo(filt, "State2");
% Uncomment the line below, and set dfdx.State1 to the
% Jacobian of the stateTransition function with respect to
% the State property of the filter object filt. Use s1idx to
% index the columns of dfdx.State1.
% % dfdx.State1(:,s1idx) =
% Uncomment the line below, and set dfdx.State2 to the
% Jacobian of the stateTransition function with respect to
% the State property of the filter object filt. Use s2idx to
% index the columns of dfdx.State2.
% % dfdx.State2(:,s2idx) =
end
end
end
% [EOF]
