Compact Maximum Correntropy-Based Error State Kalman Filter for Exoskeleton Orientation Estimation

This brief investigates the maximum correntropy-based Kalman filtering problem for exoskeleton orientation by fusing signals from accelerometers and gyroscopes. The conventional error state Kalman filter (ESKF) has been applied to many applications for orientation estimation. However, its performanc...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on control systems technology 2023-03, Vol.31 (2), p.913-920
Main Authors: Li, Shilei, Duan, Peihu, Shi, Dawei, Zou, Wulin, Duan, Pu, Shi, Ling
Format: Article
Language:English
Subjects:
Accelerometers
Algorithms
Bandwidth
Estimation
Exoskeletons
Gyroscopes
Inertial measurement units (IMUs)
Kalman filter (KF)
Kalman filters
Kernel
Measurement
non-Gaussian noise
Orientation
orientation estimation
weighted correntropy
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This brief investigates the maximum correntropy-based Kalman filtering problem for exoskeleton orientation by fusing signals from accelerometers and gyroscopes. The conventional error state Kalman filter (ESKF) has been applied to many applications for orientation estimation. However, its performance degenerates remarkably with external acceleration. In this brief, the influence of the external acceleration is analyzed and the dilemma of the conventional ESKF is declared. To address this issue, a weighted correntropy and a novel correntropy-induced metric (CIM) are provided. Then, a compact maximum correntropy-based Kalman filter (CMC-KF) is derived based on the proposed metric, which performs well both with and without non-Gaussian noises. Finally, a compact maximum correntropy-based ESKF (CMC-ESKF) is designed for orientation estimation of exoskeletons. A series of experiments are conducted to verify the effectiveness of the proposed method. Results reveal that the proposed algorithm is significantly better than the conventional ESKF and the gradient descent (GD) method, especially with external accelerations.
ISSN:1063-6536
1558-0865
DOI:10.1109/TCST.2022.3193760