Ramsete calculations

[Sumifairy]

For ramsete, what does b and zeta do and how do i calculate them? Thanks!

[maccopacco]

https://docs.wpilib.org/en/latest/docs/software/examples-tutorials/trajectory-tutorial/entering-constants.html#ramsete-parameters

[Sumifairy]

Thanks for the link, but what does those values change? Does b make the curve flatter?

[OroArmor]

b    Tuning parameter (b < 0) for which larger values make convergence more
            aggressive like a proportional term.
zeta Tuning parameter (0 < zeta < 1) for which larger values provide more damping
            in response.

[Sumifairy]

Thanks for the explanation, but what is that impact on the actual movement of the robot?

[OroArmor]

m_zeta scales out the curve, making the robot change speed slower. i cant figure out how m_b works, but this is the direct code from wpilib:

public ChassisSpeeds calculate(Pose2d currentPose,
                             Pose2d poseRef,
                             double linearVelocityRefMeters,
                             double angularVelocityRefRadiansPerSecond) {
    m_poseError = poseRef.relativeTo(currentPose);

    // Aliases for equation readability
    final double eX = m_poseError.getTranslation().getX();
    final double eY = m_poseError.getTranslation().getY();
    final double eTheta = m_poseError.getRotation().getRadians();
    final double vRef = linearVelocityRefMeters;
    final double omegaRef = angularVelocityRefRadiansPerSecond;

    double k = 2.0 * m_zeta * Math.sqrt(Math.pow(omegaRef, 2) + m_b * Math.pow(vRef, 2));

    return new ChassisSpeeds(vRef * m_poseError.getRotation().getCos() + k * eX,
                         0.0,
                         omegaRef + k * eTheta + m_b * vRef * sinc(eTheta) * eY);
}

[mustangtech]

These controller gains don’t affect the actual velocity tracking happening; they only affect how the controller outputs. Honestly, for our team, we used the constants from WPILib’s Ramsete example: https://github.com/wpilibsuite/allwpilib/blob/master/wpilibjExamples/src/main/java/edu/wpi/first/wpilibj/examples/ramsetecommand/Constants.java

    // Reasonable baseline values for a RAMSETE follower in units of meters and seconds
public static final double kRamseteB = 2;
public static final double kRamseteZeta = 0.7;

And these worked out pretty well.

[maccopacco]

With the link, @Sumifairy, I was directing you towards
Gains of 2.0 and 0.7 for b and zeta have been tested repeatedly to produce desirable results when all units were in meters.
Or as @mustangtech mentioned, using the default constants should work.

[Sumifairy]

Thanks for the replies! I have one question about their example trajectory that is supposed to go 3 meters “forward” facing 0 degrees at the end but their start is (0,0) and end is (3,0). won’t this be going to the right instead of forward? Thanks!

[Oblarg]

Headings are measured according to mathematical convention - counterclockwise from the +x axis.

[Sumifairy]

Thanks for the reply! So just to clarify, if i wanted to go forward i would change the X to positive values and side to side is a change in the Y. And then that if i wanted to face left, i would say 90, which would be facing the positive Y axis? This kinda makes sense since we just learned unit circle in trig. Thanks!

[Oblarg]

Yes, the standard is exactly the same as the unit circle you’ve learned.