Recursive KalmanNet: Deep Learning-Augmented Kalman Filtering for State Estimation with Consistent Uncertainty Quantification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mortada, Hassan, Falcon, Cyril, Kahil, Yanis, Clavaud, Mathéo, Michel, Jean-Philippe
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914032397582336
author Mortada, Hassan
Falcon, Cyril
Kahil, Yanis
Clavaud, Mathéo
Michel, Jean-Philippe
author_facet Mortada, Hassan
Falcon, Cyril
Kahil, Yanis
Clavaud, Mathéo
Michel, Jean-Philippe
contents State estimation in stochastic dynamical systems with noisy measurements is a challenge. While the Kalman filter is optimal for linear systems with independent Gaussian white noise, real-world conditions often deviate from these assumptions, prompting the rise of data-driven filtering techniques. This paper introduces Recursive KalmanNet, a Kalman-filter-informed recurrent neural network designed for accurate state estimation with consistent error covariance quantification. Our approach propagates error covariance using the recursive Joseph's formula and optimizes the Gaussian negative log-likelihood. Experiments with non-Gaussian measurement white noise demonstrate that our model outperforms both the conventional Kalman filter and an existing state-of-the-art deep learning based estimator.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11639
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Recursive KalmanNet: Deep Learning-Augmented Kalman Filtering for State Estimation with Consistent Uncertainty Quantification
Mortada, Hassan
Falcon, Cyril
Kahil, Yanis
Clavaud, Mathéo
Michel, Jean-Philippe
Signal Processing
Machine Learning
State estimation in stochastic dynamical systems with noisy measurements is a challenge. While the Kalman filter is optimal for linear systems with independent Gaussian white noise, real-world conditions often deviate from these assumptions, prompting the rise of data-driven filtering techniques. This paper introduces Recursive KalmanNet, a Kalman-filter-informed recurrent neural network designed for accurate state estimation with consistent error covariance quantification. Our approach propagates error covariance using the recursive Joseph's formula and optimizes the Gaussian negative log-likelihood. Experiments with non-Gaussian measurement white noise demonstrate that our model outperforms both the conventional Kalman filter and an existing state-of-the-art deep learning based estimator.
title Recursive KalmanNet: Deep Learning-Augmented Kalman Filtering for State Estimation with Consistent Uncertainty Quantification
topic Signal Processing
Machine Learning
url https://arxiv.org/abs/2506.11639