Struct veloren_common::comp::agent::PidController

pub struct PidController<F: Fn(Vec3<f32>, Vec3<f32>) -> f32, const NUM_SAMPLES: usize> {
    pub kp: f32,
    pub ki: f32,
    pub kd: f32,
    pub sp: Vec3<f32>,
    pv_samples: [(f64, Vec3<f32>); NUM_SAMPLES],
    pv_idx: usize,
    integral_error: f64,
    e: F,
}

PID controllers are used for automatically adapting nonlinear controls (like buoyancy for airships) to target specific outcomes (i.e. a specific height)

Fields

kp: f32

The coefficient of the proportional term

ki: f32

The coefficient of the integral term

kd: f32

The coefficient of the derivative term

sp: Vec3<f32>

The setpoint that the process has as its goal

pv_samples: [(f64, Vec3<f32>); NUM_SAMPLES]

A ring buffer of the last NUM_SAMPLES measured process variables

pv_idx: usize

The index into the ring buffer of process variables

integral_error: f64

The total integral error

e: F

The error function, to change how the difference between the setpoint and process variables are calculated

Implementations

impl<F: Fn(Vec3<f32>, Vec3<f32>) -> f32, const NUM_SAMPLES: usize> PidController<F, NUM_SAMPLES>

pub fn new(kp: f32, ki: f32, kd: f32, sp: Vec3<f32>, time: f64, e: F) -> Self

Constructs a PidController with the specified weights, setpoint, starting time, and error function

pub fn add_measurement(&mut self, time: f64, pv: Vec3<f32>)

Adds a measurement of the process variable to the ringbuffer

pub fn calc_err(&self) -> f32

The amount to set the control variable to is a weighed sum of the proportional error, the integral error, and the derivative error. https://en.wikipedia.org/wiki/PID_controller#Mathematical_form

pub fn proportional_err(&self) -> f32

The proportional error is the error function applied to the set point and the most recent process variable measurement

pub fn integral_err(&self) -> f32

The integral error is the error function integrated over all previous values, updated per point. The trapezoid rule for numerical integration was chosen because it’s fairly easy to calculate and sufficiently accurate. https://en.wikipedia.org/wiki/Trapezoidal_rule#Uniform_grid

fn update_integral_err(&mut self)

pub fn derivative_err(&self) -> f32

The derivative error is the numerical derivative of the error function based on the most recent 2 samples. Using more than 2 samples might improve the accuracy of the estimate of the derivative, but it would be an estimate of the derivative error further in the past. https://en.wikipedia.org/wiki/Numerical_differentiation#Finite_differences

Trait Implementations

impl<F: Clone + Fn(Vec3<f32>, Vec3<f32>) -> f32, const NUM_SAMPLES: usize> Clone for PidController<F, NUM_SAMPLES>

fn clone(&self) -> PidController<F, NUM_SAMPLES>

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<F: Fn(Vec3<f32>, Vec3<f32>) -> f32, const NUM_SAMPLES: usize> Debug for PidController<F, NUM_SAMPLES>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.