Debiasing 6-DOF IMU via Hierarchical Learning of Continuous Bias Dynamics

2025-04-13

AI Summary

Online bias estimation for six-degree-of-freedom IMUs remains challenging due to the time-varying nature of gyroscope and accelerometer biases. Method: This work proposes the first neural ordinary differential equation (Neural ODE) framework that explicitly models time-varying IMU biases, embedding the NODE directly into the SE(3) matrix Lie group to capture bias dynamics on the manifold. A hierarchical supervision strategy enables end-to-end training using only ground-truth poses—no bias labels required—by tightly integrating IMU preintegration with Lie-group kinematic modeling. Contribution/Results: The approach eliminates reliance on inaccessible bias ground truth. Evaluated on two public datasets and real-world experiments, it reduces pure-IMU integrated position error by 42% and decreases absolute trajectory error (ATE) of visual–inertial odometry (VIO) by 37% on average, significantly improving IMU measurement fidelity and robustness of tightly coupled VIO systems.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Biometrics, Face, Gesture & Pose
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper develops a deep learning approach to the online debiasing of IMU gyroscopes and accelerometers. Most existing methods rely on implicitly learning a bias term to compensate for raw IMU data. Explicit bias learning has recently shown its potential as a more interpretable and motion-independent alternative. However, it remains underexplored and faces challenges, particularly the need for ground truth bias data, which is rarely available. To address this, we propose a neural ordinary differential equation (NODE) framework that explicitly models continuous bias dynamics, requiring only pose ground truth, often available in datasets. This is achieved by extending the canonical NODE framework to the matrix Lie group for IMU kinematics with a hierarchical training strategy. The validation on two public datasets and one real-world experiment demonstrates significant accuracy improvements in IMU measurements, reducing errors in both pure IMU integration and visual-inertial odometry.

Problem

• Online debiasing of IMU gyroscopes and accelerometers
• Explicit modeling of continuous bias dynamics
• Reducing errors in IMU integration and odometry

Innovation

• Neural ODE framework models continuous bias dynamics
• Hierarchical training on matrix Lie group for IMU
• Uses pose ground truth instead of bias data