Error State Kalman Filter

Mar 1, 2016 1 min read
Code

The error-state Kalman filter (ESKF) is one of the tools we may use for combining IMU with magnetometer data to obtain a robust attitude estimation. It has many benefits such as avoiding issues related to over-parameterization and the consequent risk of the singularity of the involved covariance matrices. The formulation of the ESKF algorithm used for attitude estimation is as follows:

This is an image

Fig.1. Error-State Kalman Filter 1

This is an image

Fig.2. Error-State Kalman Filter 2

Below are results of the ESKF for roll, pitch, and yaw angles. The red line represents the estimation values, and the green is the ground truth.

This is an image

Fig.3. Roll

This is an image

Fig.4. Pitch

This is an image

Fig.5. Yaw
UAV Robotics Sensor Fusion IMU
Yu Zhou
Yu Zhou
Associate Scientist @Temasek Laboratories

My research interests lie in 3D visual perception and navigation, applied machine learning.

Related