ADUGS-VINS: Generalized Visual-Inertial Odometry for Robust Navigation in Highly Dynamic and Complex Environments

TL;DR

ADUGS-VINS tackles the challenge of accurate visual-inertial odometry in highly dynamic and partially occluded environments. It fuses an enhanced SORT tracker with a promptable segmentation model (Mobile SAM) to robustly identify and exclude dynamic regions, while employing an adaptive Kalman-filter-based update and multi-category segmentation to maintain reliable pose estimates. The method further uses ORB features with ANMS and a compensation strategy to preserve sufficient static features for stable optimization, yielding strong results across VIODERN45, OpenLORIS, VIODE, and real-world tests, often outperforming state-of-the-art baselines. This work advances practical navigation in dynamic scenes, with implications for robotics, drones, and autonomous systems requiring robust VIO under complex object motion and occlusions.

Abstract

Visual-inertial odometry (VIO) is widely used in various fields, such as robots, drones, and autonomous vehicles. However, real-world scenes often feature dynamic objects, compromising the accuracy of VIO. The diversity and partial occlusion of these objects present a tough challenge for existing dynamic VIO methods. To tackle this challenge, we introduce ADUGS-VINS, which integrates an enhanced SORT algorithm along with a promptable foundation model into VIO, thereby improving pose estimation accuracy in environments with diverse dynamic objects and frequent occlusions. We evaluated our proposed method using multiple public datasets representing various scenes, as well as in a real-world scenario involving diverse dynamic objects. The experimental results demonstrate that our proposed method performs impressively in multiple scenarios, outperforming other state-of-the-art methods. This highlights its remarkable generalization and adaptability in diverse dynamic environments, showcasing its potential to handle various dynamic objects in practical applications.

Figures (9)

• Figure 1: Overview of ADUGS-VINS. Our method effectively segments various moving objects even under conditions of partial occlusion, demonstrating the efficacy of ADUGS-VINS in segmentation within complex environments.
• Figure 2: Changes of measurement noise covariance matrix updating function w.r.t. parameter $\lambda$.
• Figure 3: Feature points distribution of ADUGS-VINS in VIODE dataset.The truck in \ref{['feature(a)']} exhibits a partial occlusion relationship. A significant portion of the car in \ref{['feature(b)']} is outside the field of camera view.
• Figure 4: The heatmap illustrates RMSE ATE in relation to the maximum feature points $N_{max}$ and minimum feature point distance $D_{min}$ for the "high" sequence within the parking lot environment of the VIODE dataset. The color bar represents the range of ATE values.
• Figure 5: A comparison of the trajectories of various VIO system on the "high" sequence of the VIODE dataset is presented. The ground truth is highlighted with a dashed yellow line. It can be observed that the trajectory obtained by ADUGS-VINS is the most accurate, with a high degree of overlap with the ground truth, reflecting the precision of the algorithm.