KalmanNet: Neural Network Aided Kalman Filtering for Partially Known Dynamics

TL;DR

KalmanNet is a neural network-enhanced Kalman filter that learns to estimate states in systems with non-linear dynamics and partial information, improving accuracy over traditional methods.

Contribution

It introduces a hybrid approach combining Kalman filtering structure with neural networks to handle non-linearities and model mismatch.

Findings

KalmanNet outperforms classic filters in non-linear scenarios.

It effectively handles model mismatch and partial knowledge.

The method maintains data efficiency and interpretability.

Abstract

State estimation of dynamical systems in real-time is a fundamental task in signal processing. For systems that are well-represented by a fully known linear Gaussian state space (SS) model, the celebrated Kalman filter (KF) is a low complexity optimal solution. However, both linearity of the underlying SS model and accurate knowledge of it are often not encountered in practice. Here, we present KalmanNet, a real-time state estimator that learns from data to carry out Kalman filtering under non-linear dynamics with partial information. By incorporating the structural SS model with a dedicated recurrent neural network module in the flow of the KF, we retain data efficiency and interpretability of the classic algorithm while implicitly learning complex dynamics from data. We demonstrate numerically that KalmanNet overcomes non-linearities and model mismatch, outperforming classic filtering methods operating with both mismatched and accurate domain knowledge.