BLENDS: Bayesian Learning-Enhanced Deep Smoothing for GNSS-Denied Environments

2026-06-21

AI Summary

This work addresses the challenge of maintaining high-precision navigation with low-cost inertial systems during GNSS signal outages. The authors propose a novel framework that synergistically integrates Transformer-based deep learning with Bayesian smoothing, achieving, for the first time, joint optimization between data-driven modeling and Bayesian estimation. The approach adaptively tunes process and observation noise covariances in a data-driven manner and corrects state estimates to automatically compensate for systematic GNSS biases. By reducing reliance on measurement accuracy inherent in conventional smoothing methods, the proposed method significantly lowers position root-mean-square error—by up to 25.6%—on a quadrotor GNSS-denied dataset, markedly decreasing estimation uncertainty and surpassing the navigation precision achievable with standard GNSS measurements alone.

Abstract

Maintaining accurate navigation during GNSS outages remains a significant challenge for autonomous systems relying on low-cost inertial sensors. While classical smoothing methods, such as the two-filter smoother and Rauch-Tung-Striebel smoother, exploit measurements collected before and after an outage, their performance remains limited by the accuracy of conventional GNSS measurements. This paper presents Bayesian learning-enhanced navigation with deep smoothing (BLENDS), a transformer-based framework that augments Bayesian smoothing with learned covariance adaptation and state correction. The proposed method preserves the statistical foundations of Bayesian estimation while leveraging data-driven learning to improve navigation accuracy. Evaluations on the quadrotor dataset with GNSS outages demonstrate that BLENDS consistently outperforms both model-based smoothers, achieving up to 25.6% improvement in the position root mean square error while also reducing estimation uncertainty. Furthermore, BLENDS learns to compensate for the systematic bias between conventional GNSS positioning and RTK ground truth, enabling accuracy beyond that achievable with conventional GNSS measurements alone. The results demonstrate the potential of learning-enhanced Bayesian smoothing for resilient and high-accuracy navigation in GNSS-challenged environments.

Problem

• GNSS-denied navigation
• inertial navigation
• smoothing
• position accuracy
• autonomous systems

Innovation

• Bayesian smoothing
• deep learning
• transformer
• GNSS-denied navigation
• covariance adaptation