KalmanNet: Neural Network Aided Kalman Filtering for Partially Known Dynamics
This addresses real-time state estimation problems in signal processing for systems with non-linear or partially known dynamics, representing an incremental improvement over traditional Kalman filtering.
The authors tackled state estimation under non-linear dynamics with partial model knowledge by introducing KalmanNet, a neural network-aided Kalman filter that learns from data, and demonstrated it outperforms classic filtering methods in numerical tests.
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.