计算机与现代化Issue(4)：104-110,7.DOI:10.3969/j.issn.1006-2475.2026.04.014
灭点辅助的机器人位姿优化方法
Robot Pose Optimization Using Vanishing Point Constraints
摘要
Abstract
The classical PL-VINS point and line system utilizes a mixture of point features and structural line features to achieve localization in environments with scarce visual features.This approach can obtain a more accurate robot position to a certain ex-tent and improve the system's localization accuracy.However,the system ignores structural information such as the parallelism and orthogonality of the line features,and thus fails to fully utilize the extracted line feature information,leaving room for further improvement in localization accuracy.To obtain a more accurate robot position,a robot position optimization method based on vanishing point assistance is proposed.First,the definition and properties of the vanishing point are briefly introduced.Then the computation of the vanishing point observation and estimation values is investigated in detail.A hybrid strategy combining sam-pling and searching is proposed to calculate the vanishing point observation value.Meanwhile,the estimation accuracy of vanish-ing points is improved by utilizing the transformation relationship between the coordinate systems and the projection relationship.Subsequently,residuals are constructed based on both vanishing point observations and estimation values.Theses residuals are jointly optimized using the Gaussian Newton nonlinear optimization method to obtain a more accurate robot position,thereby en-hancing the overall localization performance of the system.Simulation and experimental results show that the vanishing point re-sidual factor constructed based on the vanishing point observation value obtained by the hybrid strategy and the vanishing point estimation value obtained based on the spatial transformation relationship effectively improves the localization accuracy of the sys-tem after the joint optimization.Specifically,the method achieves an average accuracy improvement of over 8%after joint optimi-zation.关键词
机器人/灭点/定位/精度Key words
robot/vanishing point/localization/precision分类
信息技术与安全科学引用本文复制引用
陈阵,胡为..灭点辅助的机器人位姿优化方法[J].计算机与现代化,2026,(4):104-110,7.基金项目
辽宁省自然科学基金资助项目(2022JH2/101300283) （2022JH2/101300283）
辽宁省教育厅高校科研基金资助项目(JYT2020074) （JYT2020074）
