Physics constrained learning of stochastic characteristics

2025-07-16

AI Summary

In vehicle state estimation, conventional filters often diverge due to inaccurate manual tuning of process and measurement noise covariances, especially when true noise statistics are unknown. Method: This paper proposes a physics-constrained joint noise covariance identification framework. It constructs a multi-objective loss function based on innovation sequences and jointly optimizes process and observation noise covariance matrices in an end-to-end manner, integrating optimization algorithms with deep learning under dynamical model constraints. Contribution/Results: Departing from heuristic parameter tuning, the method is the first to embed differentiable physical constraints directly into the noise modeling pipeline, enabling real-time, robust online covariance estimation. Experiments on real-world automotive datasets demonstrate substantial improvements in estimation accuracy and stability for UKF and ESKF filters, effectively preventing filter divergence caused by erroneous noise assumptions.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Accurate state estimation requires careful consideration of uncertainty surrounding the process and measurement models; these characteristics are usually not well-known and need an experienced designer to select the covariance matrices. An error in the selection of covariance matrices could impact the accuracy of the estimation algorithm and may sometimes cause the filter to diverge. Identifying noise characteristics has long been a challenging problem due to uncertainty surrounding noise sources and difficulties in systematic noise modeling. Most existing approaches try identifying unknown covariance matrices through an optimization algorithm involving innovation sequences. In recent years, learning approaches have been utilized to determine the stochastic characteristics of process and measurement models. We present a learning-based methodology with different loss functions to identify noise characteristics and test these approaches' performance for real-time vehicle state estimation

Problem

• Identify noise characteristics in state estimation
• Overcome uncertainty in process and measurement models
• Improve accuracy of real-time vehicle state estimation

Innovation

• Physics constrained learning for stochastic characteristics
• Learning-based methodology with loss functions
• Real-time vehicle state estimation testing