insCFMotionPose

Motion model for 3-D pose estimation using insCF

Since R2026a

expand all in page

    Description

    The insCFMotionPose object models 3-D pose (orientation and position) assuming a constant angular velocity and constant acceleration. Passing an insCFMotionPose object to an insCF object enables the estimation of 3-D pose, including orientation, position, and velocity.

    Creation

    Syntax

    model = insCFMotionPose

    Description

    model = insCFMotionPose creates an insCFMotionPose object. Passing the created model to an insCF object enables the insCF object to estimate the following quantities as parts of the state.

    • The orientation quaternion from the navigation frame to the body frame.

    • The position of the platform, expressed in the navigation frame.

    • The velocity of the platform, expressed in the navigation frame.

    To estimate pose and velocity using an insCFMotionPose object with an insCF filter, you must pass this data to the insCF object:

    • Gyroscope data for predicting the orientation.

    • Data from either of the two orientation-correcting sensors, accelerometer or magnetometer, for correcting orientation.

    • Accelerometer data for predicting position and velocity.

    • GPS data for correcting position.

    example

    Examples

    collapse all

    Open Live Script

    Estimate pose (orientation and position) and velocity from IMU and GPS data by using a complementary filter, represented by an insCF filter object.

    Load the racetrackINSCFDataset.mat file, which contains a timetable with simulated accelerometer, gyroscope, magnetometer, and GPS sensor data for a racetrack-like trajectory, into the workspace. The file also contains the sample rate of the data.

    load("raceTrackINSCFDataset.mat")

    Create an insCF filter object that includes the accelerometer, gyroscope, magnetometer, and GPS sensor objects, as well as an insCFMotionPose motion model object for modeling pose.

    % Sensor object for predicting orientation
gyro = insCFGyroscope;
% Sensor objects for correcting orientation
accel = insCFAccelerometer;
mag = insCFMagnetometer;
% Sensor object for correcting position
gps = insCFGPS(ReferenceLocation=localOrigin);

% Pose motion model
motModel = insCFMotionPose;

% Filter object
filt = insCF(accel,mag,gyro,gps,motModel);

    Specify the initial position, velocity, and orientation from the measurement data.

    % Set initial position
stateparts(filt,"Position",initPos)
% Set initial orientation
stateparts(filt,"Orientation",initOrient)
% Set initial velocity
stateparts(filt,"Velocity",initVel)

    Specify the sensor gains.

    % Set Accelerometer gain
gainparts(filt,"Accelerometer",9.6335e-09)
% Set Magnetometer gain
gainparts(filt,"Magnetometer",0.01)
% Set GPS gain
gainparts(filt,"GPS",0.039246)

    Fuse the sensor data using the filter. The filter estimates pose (orientation and position) and velocity. Extract the estimated position.

    outTT = estimateStates(filt,sensorData);
estPos = outTT.Position;

    Calculate the position RMS error.

    posErr = truePos - estPos; 
pRMS = sqrt(mean(posErr.^2));
fprintf("Position RMS Error:\tX: %.3f, Y: %.3f, Z: %.3f (meters) \n", pRMS(1),pRMS(2),pRMS(3))
    Position RMS Error:	X: 0.372, Y: 0.493, Z: 0.303 (meters)

    Visualize the estimated pose against the ground truth and GPS data.

    figure
plot(truePos(:,1),truePos(:,2),".")
hold on
plot(estPos(:,1),estPos(:,2),"r.-")
gpsLocal = lla2ned(sensorData.GPS,localOrigin,"flat");
plot(gpsLocal(:,1),gpsLocal(:,2),".")

title("Pose Estimate")
legend("Ground Truth","Estimated","GPS")

grid on
xlabel("N (m)")
ylabel("E (m)")
axis equal

    Figure contains an axes object. The axes object with title Pose Estimate, xlabel N (m), ylabel E (m) contains 3 objects of type line. One or more of the lines displays its values using only markers These objects represent Ground Truth, Estimated, GPS.

    Extended Capabilities

    expand all

    C/C++ Code Generation
    Generate C and C++ code using MATLAB® Coder™.

    Version History

    Introduced in R2026a

    See Also

    Objects