【摘要】 在视觉SLAM系统中,闭环检测是非常重要的一个模块,它主要用于解决移动机器人在位置估计时随时间漂移的问题。移动机器人搭载相机在场景中运动,通过前端视觉里程计估计自身位姿和场景地图点坐标。视觉里程计历史时刻的估计误差会传递到下一时刻,导致一段时间后估计的结果出现累计误差,闭环检测可有效减少该累计误差。针对传统基于点特征的视觉SLAM闭环检测在点稀缺环境下精度和鲁棒性较差的问题,提出结合环境权重点线特征的闭环检测系统。该系统相比传统的基于点特征的闭环检测算法,系统鲁棒性更强,检测精确度更高。更多还原

【Abstract】 In the visual SLAM system,loop closure detection is a very important module,which is mainly used to solve the problem that the mobile robot drifts with time during localization estimation. The mobile robot is equipped with a camera to move in the scene,and its own positions and scene map point coordinates are estimated by the front-end visual odometry. The estimated error of the history visual odometry is passed to the next moment,resulting in a cumulative error in the estimated result over time. Aiming at the problem that the traditional loop closure detection based on point features is less robust in low-texture environment,a system based on the weight line features of environment is proposed. Comparing with the traditional loop closure detection algorithm based on point feature,the system has stronger robustness and higher detection accuracy.更多还原