A Visual-Inertial Motion Prior SLAM for Dynamic Environments

2025-03-30

AI Summary

To address the failure of conventional visual-inertial SLAM (VI-SLAM) in dynamic environments due to moving landmarks, this paper proposes a robust tightly coupled visual-inertial SLAM framework. Our method explicitly models inertial motion priors as landmark reprojection error minimization, enabling a probabilistic pre-filtering mechanism for dynamic features—the first such formulation. It further integrates epipolar constraints for coarse dynamic feature rejection and introduces a novel bundle adjustment (BA) residual model within the sliding-window nonlinear optimization, specifically designed to suppress dynamic outliers. Extensive evaluations on multiple dynamic-scene datasets demonstrate that our approach achieves 12.6% higher localization accuracy and 19.3% lower computational latency compared to state-of-the-art methods including VINS-Mono and EKLT.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Motion & Tracking
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The Visual-Inertial Simultaneous Localization and Mapping (VI-SLAM) algorithms which are mostly based on static assumption are widely used in fields such as robotics, UAVs, VR, and autonomous driving. To overcome the localization risks caused by dynamic landmarks in most VI-SLAM systems, a robust visual-inertial motion prior SLAM system, named (IDY-VINS), is proposed in this paper which effectively handles dynamic landmarks using inertial motion prior for dynamic environments to varying degrees. Specifically, potential dynamic landmarks are preprocessed during the feature tracking phase by the probabilistic model of landmarks' minimum projection errors which are obtained from inertial motion prior and epipolar constraint. Subsequently, a bundle adjustment (BA) residual is proposed considering the minimum projection error prior for dynamic candidate landmarks. This residual is integrated into a sliding window based nonlinear optimization process to estimate camera poses, IMU states and landmark positions while minimizing the impact of dynamic candidate landmarks that deviate from the motion prior. Finally, experimental results demonstrate that our proposed system outperforms state-of-the-art methods in terms of localization accuracy and time cost by robustly mitigating the influence of dynamic landmarks.

Problem

• Overcoming localization risks from dynamic landmarks in VI-SLAM
• Handling dynamic environments using inertial motion prior
• Improving localization accuracy and time cost in dynamic settings

Innovation

• Uses inertial motion prior for dynamic landmarks
• Preprocesses dynamic landmarks with probabilistic model
• Integrates BA residual in nonlinear optimization