Scribd
Upload
91 views

Module 2 Lesson 4: An Improved Ekf: The Error-State Extended Kalman

The document describes the error-state extended Kalman filter (ES-EKF). The ES-EKF separates the state into a nominal state and error state. It estimates the error state directly using linearization and uses it to correct the nominal state. This can provide better performance than the standard EKF. The ES-EKF also provides a natural way to handle states with constraints, like rotations in 3D.

Uploaded by

Mary Justice
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
91 views

Module 2 Lesson 4: An Improved Ekf: The Error-State Extended Kalman

The document describes the error-state extended Kalman filter (ES-EKF). The ES-EKF separates the state into a nominal state and error state. It estimates the error state directly using linearization and uses it to correct the nominal state. This can provide better performance than the standard EKF. The ES-EKF also provides a natural way to handle states with constraints, like rotations in 3D.

Uploaded by

Mary Justice
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

MODULE 2 LESSON 4

AN IMPROVED EKF: THE ERROR-


STATE EXTENDED KALMAN
The Error-State EKF (ES-EKF)
By the end of this video, you will be able to


• Describe the error-state formulation of the Extended Kalman Filter


• Describe the advantages of the Error-state EKF over the vanilla EKF
What’s in a State?
We can think of the vehicle state as composed of two
parts: x = x̂ + δx

True State
 Nominal State Error State


(“Large”) (“Small”)

Position

Error state δx
• We can continuously update the
nominal state by integrating the
True state x motion model

Nominal state x̂
Time
• Modelling errors and process noise
accumulate into the error state
The Error-State Extended Kalman Filter
The Error-State Extended Kalman Filter estimates the error state directly and
uses it as a correction to the nominal state:

Linearized motion model Linearized measurement model

xk = fk−1(x̂k−1, uk−1, 0) + Fk−1 (xk−1 − x̂k−1) + Lk−1wk−1 yk = hk(x̌k, 0) + Hk (xk − x̌k) + Mkvk

xk − fk−1(x̂k−1, uk−1, 0) = Fk−1 (xk−1 − x̂k−1) + Lk−1wk−1 yk = hk(x̌k, 0) + Hk (xk − x̌k) + Mkvk
δxk δxk−1 δxk

Error state Error state


The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model

x̌k = fk−1(xk−1, uk−1, 0)

This could be
x̌ or x̂
k−1 k−1
The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model
2. Propagate uncertainty
T T
P̌k = Fk−1Pk−1Fk−1 + Lk−1Qk−1Lk−1

This could be
P̌ or P̂
k−1 k−1
The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model
2. Propagate uncertainty
3. If a measurement is available:
1. Compute Kalman Gain
Kk = P̌kHTk (HkP̌kHTk + R)−1
The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model
2. Propagate uncertainty
3. If a measurement is available:
1. Compute Kalman Gain
2. Compute error state
δx̂k = Kk(yk − hk(x̌k, 0))
The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model
2. Propagate uncertainty
3. If a measurement is available:
1. Compute Kalman Gain
2. Compute error state
3. Correct nominal state
x̂k = x̌k + δx̂k
The Error-State Extended Kalman Filter
Loop:
1. Update nominal state with motion model
2. Propagate uncertainty
3. If a measurement is available:
1. Compute Kalman Gain
2. Compute error state
3. Correct nominal state
4. Correct state covariance
P̂ k = (1 − KkHk)P̌k
Why Use the ES-EKF?

1. Better performance compared to the vanilla EKF



The “small” error state is more amenable to linear filtering than the “large”
nominal state, which can be integrated nonlinearly

2. Easy to work with constrained quantities (e.g., rotations in 3D)



We can also break down the state using a generalized composition operator

x = x̂ ⊕ δx

True State
 Nominal State Error State


(Overparametrized, (Minimal parametrization,
unconstrained)
constrained)
Summary | The Error-State EKF (ES-EKF)
• The error-state formulation separates the state into a “large” nominal
state and a “small” error state.


• The ES-EKF uses local linearization to estimate the error state and uses it
to correct the nominal state.


• The ES-EKF can perform better than the vanilla EKF, and provides a
natural way to handle constrained quantities like rotations in 3D.

You might also like