基于滑窗非线性优化的双目视觉SLAM算法  

Sliding window based binocular SLAM using nonlinear optimization

作　　者：毛家宁 赵龙[1] MAO Jianing;ZHAO Long(School of Automation Science and Electrical Engineering,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191

出　　处：《应用科技》2020年第1期55-60,共6页Applied Science and Technology

基　　金：国家自然科学基金项目(41874034,41574024);国家重点研发计划项目(2016YFB0502102);北京市自然科学基金项目(4162035);中国航空基础科学基金项目(2016ZC51024).

摘　　要：针对双目基于ORB特征的即时定位与地图构建(ORB-SLAM)算法位姿估计精度较低的问题,对其进行了改进,提出了一种基于滑窗非线性优化的双目视觉SLAM算法。该算法引入了滑窗思想,在状态估计过程中对最新多帧图像对应系统的状态进行了解算;重构了位姿估计过程中的代价函数,通过高斯牛顿法优化状态变量;采用边缘化策略,有效限制了待求解状态量的数目,并将其余状态量及与其相关的视觉测量转换为求解当前滑窗内状态的先验约束。通过在公开数据集上测试,验证了算法的精确性和有效性,实验结果表明:该算法能有效估计相机位姿且不发生显著漂移,与双目ORBSLAM算法相比,位姿估计精度有了显著提高。To solve the low pose estimation accuracy problem of binocular ORB-SLAM(oriented fast and rotated BRIEF-simultaneous localization and mapping)algorithm,we propose an improved binocular vision SLAM algorithm based on sliding window nonlinear optimization.Introducing the idea of sliding window,the algorithm solve the state corresponding to the latest multi-frame image.The cost function used in pose estimation is reconstructed,and the state variables are optimized by Gauss-Newton method.The marginalization strategy is adopted to effectively limit the number of state variables to be solved,and the remaining state variables and their related visual measurements are converted into prior constraints to solve the current state in the sliding window.The accuracy and effectiveness of the proposed algorithm are verified by the test on the open data set.The experimental results show that the proposed algorithm can effectively estimate camera pose without significant drift,and the accuracy of pose estimation is significantly improved compared with the binocular ORB-SLAM algorithm.

关 键 词：双目视觉 图像处理 导航定位 特征法 地图构建 非线性优化 滑窗 边缘化 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]