SchurVINS: Schur Complement-Based Lightweight Visual Inertial Navigation System

TL;DR

SchurVINS addresses the challenge of achieving high-accuracy visual-inertial navigation on resource-limited devices by introducing an EKF-based framework that uses a complete residual model and Schur complement to separate ego-motion and landmark updates. The method builds an equivalent residual comprising gradient, Hessian, and observation covariance, then decomposes the problem into pose and landmark subproblems, with an EKF-based landmark solver handling landmark estimation within a sliding window. Experiments on EuRoC and TUM-VI show SchurVINS achieves state-of-the-art efficiency among EKF-based methods while maintaining competitive accuracy compared with optimization-based VINS, and ablation studies confirm the effectiveness of the landmark solver. The work enables accurate, real-time VINS on devices with limited compute by exploiting sparsity and structured residuals, with potential for further refinement via local map improvements.

Abstract

Accuracy and computational efficiency are the most important metrics to Visual Inertial Navigation System (VINS). The existing VINS algorithms with either high accuracy or low computational complexity, are difficult to provide the high precision localization in resource-constrained devices. To this end, we propose a novel filter-based VINS framework named SchurVINS, which could guarantee both high accuracy by building a complete residual model and low computational complexity with Schur complement. Technically, we first formulate the full residual model where Gradient, Hessian and observation covariance are explicitly modeled. Then Schur complement is employed to decompose the full model into ego-motion residual model and landmark residual model. Finally, Extended Kalman Filter (EKF) update is implemented in these two models with high efficiency. Experiments on EuRoC and TUM-VI datasets show that our method notably outperforms state-of-the-art (SOTA) methods in both accuracy and computational complexity. The experimental code of SchurVINS is available at https://github.com/bytedance/SchurVINS.

Figures (4)

• Figure 1: Comparison of run time, CPU usage and RMSE evaluated on EuRoC dataset. Different shapes and colors indicate different methods and precision, respectively.
• Figure 2: Framework of SchurVINS, which shows the relationship between SVO and SchurVINS. $P_1$ to $P_m$ represent the valid landmarks of the surrounding environment which are employed to construct residual model.
• Figure 3: A schematic of our system for ten landmarks and the sliding window of size three shown in (a), and the Hessian or Covariance of different methods shown in (b)-(d). (b) shows our algorithm in which the covariance of every single landmark is independent from the entire covariance of poses in the sliding window. (c) demonstrates the Hessian of both landmarks and poses in the sliding window. (d) demonstrates traditional hybrid MSCKF with the Covariance of both selected landmarks and poses in the sliding window.
• Figure 4: The experimental trajectory and point cloud of SchurVINS on TUM-VI and EuRoC datasets.