# MUSE: Multimodal Uncertainty Quantification of State Estimation

Source: [MUSE: Multimodal Uncertainty Quantification of State Estimation](<https://www.sota2.com/research/paper/6a0bee9fdbc836b87d82c4ae>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-20T05:00:13.583Z

Authors: Minkyung Kim, Henry Che, Bhargav Chandaka, Bhumsitt Pramuanpornsatid, Chengyu Yang, Sheng Cheng, Xiaofeng Wang, Naira Hovakimyan, Shenlong Wang

Published: 2026-05-17T12:33:02.000Z

## Abstract

Accurate visual state estimation has been a central topic in robotics with a wide range of applications in robot navigation, autonomous driving, and autonomous flight. Recent advances in robot perception have led to significant improvements in the accuracy and robustness of state estimation, yet a fundamental challenge remains in how to quantify and calibrate its precision, i.e., how confident we are in an estimate and whether failures can be detected. This issue is particularly pronounced in visual-inertial odometry (VIO), where the heteroscedastic and multimodal nature of the problem makes uncertainty quantification especially difficult. This paper introduces MUSE (Multimodal Uncertainty Quantification of State Estimation), a novel real-time learning-based framework that leverages the strong and efficient sequential modeling capacity of Mamba to estimate localization uncertainty from multiple asynchronous sensor streams. Experiments on both public and in-house datasets demonstrate that MUSE achieves superior reliability and robustness compared to existing uncertainty quantification methods, and ablation studies justify the benefits of its key design choices.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Uncertainty Quantification on EuRoC MSCKF](<https://www.sota2.com/research/sota/uncertainty-quantification-on-euroc-msckf.md>): Uncertainty Quantification; EuRoC MSCKF; LL: 44.33.

- [Pose Correction on UnCal-Flight (Yaw Constant)](<https://www.sota2.com/research/sota/pose-correction-on-uncal-flight-yaw-constant.md>): Pose Correction; UnCal-Flight (Yaw Constant); RMSE (m): 0.035.

- [Pose Correction on UnCal-Flight Yaw Forward](<https://www.sota2.com/research/sota/pose-correction-on-uncal-flight-yaw-forward.md>): Pose Correction; UnCal-Flight Yaw Forward; RMSE (m): 0.04.

- [Uncertainty Quantification on UnCal-Flight Yaw Constant](<https://www.sota2.com/research/sota/uncertainty-quantification-on-uncal-flight-yaw-constant.md>): Uncertainty Quantification; UnCal-Flight (Yaw Constant); LL: 32.

- [Uncertainty Quantification on UnCal-Flight (Yaw Forward)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-uncal-flight-yaw-forward.md>): Uncertainty Quantification; UnCal-Flight Yaw Forward; Log Likelihood (LL): 29.46.

- [Pose Correction on EuRoC](<https://www.sota2.com/research/sota/pose-correction-on-euroc.md>): Pose Correction; EuRoC; RMSE: 0.028.

- [Uncertainty Quantification on EuRoC ORB-SLAM 3](<https://www.sota2.com/research/sota/uncertainty-quantification-on-euroc-orb-slam-3.md>): Uncertainty Quantification; EuRoC ORB-SLAM 3; Log-Likelihood: 54.83.

- [Uncertainty Quantification on EuRoC VINS-Fusion](<https://www.sota2.com/research/sota/uncertainty-quantification-on-euroc-vins-fusion.md>): Uncertainty Quantification; EuRoC VINS-Fusion; LL: 37.95.

- [Uncertainty Quantification on EuRoC DPVO](<https://www.sota2.com/research/sota/uncertainty-quantification-on-euroc-dpvo.md>): Uncertainty Quantification; EuRoC DPVO; LL: 49.06.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.17421>)

[PDF](<https://arxiv.org/pdf/2605.17421.pdf>)