一种基于点线特征协同的高精度视觉惯性系统OA

Simultaneous localization and mapping(SLAM)is a fundamental technology that enables intelligent agents,such as robots and unmanned aerial vehicles(UAVs),to achieve autonomous navigation and environmental perception in unknown environments.However,current mainstream visual-inertial systems(VINS)heavily rely on point features within images,leading to a sharp degradation in localization accuracy when operating in environments with sparse textures or abundant repetitive structures.To address this issue,a high-precision visual-inertial navigation system is proposed based on the collaboration of point and line features.In the frontend,while retaining the efficient KLT optical flow algorithm for tracking corner features,we introduce a novel dynamic line feature extraction strategy assisted by motion priors,which significantly enhances the efficiency and robustness of line feature tracking.Subsequently,in the backend,we formulate a 3D line feature re-projection error factor based on Plücker coordinates and design a corresponding local parameterization method to ensure the numerical validity of the non-linear optimization.The improved algorithm is evaluated on the public EuRoC dataset.Experimental results demonstrate that,the proposed method reduces the root mean square error(RMSE)by 10.0%and 4.9%relative to VINS-Mono and PL-VINS respectively,effectively validating the efficacy of the proposed algorithm.