Swerve Steering and Driving Controls Lesson

Swerve Steering and Driving Controls

Kinematics tells a swerve module what to do (e.g., "go to 45 degrees"). The control system figures out how to do it. For this, we use two powerful but different strategies: PID control for steering and Feedforward control for driving.

Steering Control: Hitting the Perfect Angle with PID

The goal of steering is to turn the wheel to an exact angle as quickly and accurately as possible. A PID Controller is a feedback loop that constantly looks at the difference between where the wheel is and where it wants to be and adjusts its output accordingly.

The Three Components of PID

P (Proportional): The main driving force. The bigger the error (difference between target and current angle), the more power it applies.
I (Integral): Corrects for small, steady errors. It accumulates error over time to overcome things like friction and ensure perfect precision.
D (Derivative): The "brake." It looks at how fast the error is changing to slow the module down as it approaches the target, preventing overshoot.

// In the SwerveModule class...
private final PIDController turningPidController;

public SwerveModule(...) {
    turningPidController = new PIDController(kP, kI, kD);
    
    // This is crucial for swerve: it makes the PID controller understand circular movement (like a compass).
    // Turning from 359° to 1° is a small 2° turn, not a huge 358° turn.
    turningPidController.enableContinuousInput(-Math.PI, Math.PI);
}

public void setDesiredState(SwerveModuleState desiredState) {
    // Use the PID controller to calculate the output for the turning motor.
    double pidOutput = turningPidController.calculate(getTurningAngle(), desiredState.angle.getRadians());
    turningMotor.set(pidOutput);
    // ... drive motor logic ...
}

Driving Control: Predicting the Perfect Speed with Feedforward

For the drive wheels, we use a more predictive method: a Feedforward Controller. While PID is reactive (it corrects errors after they happen), Feedforward is predictive. It uses a mathematical model of the motor to predict the exact voltage needed to achieve a desired speed without waiting for an error.

The Fly Ball Analogy

Think of an outfielder catching a fly ball. Based on the initial trajectory, they immediately run to a spot where they predict the ball will land (Feedforward). While running, they look up to make small corrections (Feedback/PID). Feedforward does 95% of the work upfront.

WPILib's `SimpleMotorFeedforward` class uses a few key constants (found using the SysId tool) to model our motor: `kS` (to overcome static friction), `kV` (to hold velocity), and `kA` (to accelerate).

// In the SwerveModule class...
private final SimpleMotorFeedforward driveFeedforward;

public SwerveModule(...) {
    driveFeedforward = new SimpleMotorFeedforward(kS, kV, kA);
}

public void setDesiredState(SwerveModuleState desiredState) {
    // Use the feedforward controller to predict the voltage for the drive motor.
    double ffOutput = driveFeedforward.calculate(desiredState.speedMetersPerSecond);
    driveMotor.setVoltage(ffOutput);
    // ... turning motor logic ...
}

Test Your Knowledge

Question: Which control strategy is "predictive" and uses a physical model of the motor to calculate the necessary output, and which is "reactive," constantly correcting for errors?