基于VSLAM的自主移动机器人三维同时定位与地图构建

发布时间：2018-03-13 13:52

本文选题：计算机视觉　切入点：同时定位与地图构建　出处：《计算机应用》2017年10期 　论文类型：期刊论文

【摘要】：移动机器人在探索未知环境且没有外部参考系统的情况下,面临着同时定位和地图构建(SLAM)问题。针对基于特征的视觉SLAM(VSLAM)算法构建的稀疏地图不利于机器人应用的问题,提出一种基于八叉树结构的高效、紧凑的地图构建算法。首先,根据关键帧的位姿和深度数据,构建图像对应场景的点云地图;然后利用八叉树地图技术进行处理,构建出了适合于机器人应用的地图。将所提算法同RGB-D SLAM(RGB-Depth SLAM)算法、Elastic Fusion算法和ORB-SLAM(Oriented FAST and Rotated BRIEF SLAM)算法通过权威数据集进行了对比实验,实验结果表明,所提算法具有较高的有效性、精度和鲁棒性。最后,搭建了自主移动机器人,将改进的VSLAM系统应用到移动机器人中,能够实时地完成自主避障和三维地图构建,解决稀疏地图无法用于避障和导航的问题。
[Abstract]:The mobile robot is faced with the problem of simultaneous localization and map construction without an external reference system while exploring unknown environment. The sparse map constructed by the feature-based visual SLAM- VSLAM-based algorithm is not conducive to the robot application. An efficient and compact map construction algorithm based on octree structure is proposed. Firstly, based on the position and depth data of key frames, the point cloud map of the corresponding scene is constructed, and then the octree map technology is used to process the map. A map suitable for robot applications is constructed. The proposed algorithm is compared with RGB-D SLAM(RGB-Depth slam algorithm and ORB-SLAM(Oriented FAST and Rotated BRIEF algorithm through authoritative data sets. The experimental results show that the proposed algorithm is highly effective. Finally, the autonomous mobile robot is built, and the improved VSLAM system is applied to the mobile robot, which can complete the autonomous obstacle avoidance and 3D map construction in real time, and solve the problem that sparse map can not be used for obstacle avoidance and navigation.
【作者单位】：装甲兵工程学院控制工程系;江西理工大学信息工程学院;
【基金】：国家自然科学基金资助项目(61663014)~~
【分类号】：TP242

【相似文献】