摘要
同步定位与建图(SLAM)是移动机器人在复杂环境下进行环境感知的重要手段。针对传统的SLAM算法缺乏语义信息的问题,文章使用基于深度学习的语义分割算法,利用轻量化的DeepLabV3+模型在动态环境下进行实时语义分割,得到二维语义标签。通过VINS-Mono算法推测相机位姿,同时结合深度数据、语义信息生成三维语义点云,并将点云转化成八叉树地图进行表示。实验结果表明,文章提出的算法可以满足实时情况下构建语义地图的要求。
Simultaneous Localization and Mapping(SLAM)is an important means for mobile robots to perceive the environment in complex environments.Aiming at the problem that traditional SLAM algorithm lacks semantic information,this paper uses a semantic segmentation algorithm based on deep learning,and uses a lightweight DeepLabV3+model to perform real-time semantic segmentation in a dynamic environment to obtain two-dimensional semantic tags.The camera pose is inferred through the VINS-Mono algorithm,and the 3D semantic point cloud is generated by combining the depth data and semantic information.The point cloud is converted into an octree map for representation.The experimental results show that the algorithm proposed in this paper can meet the requirements of building semantic maps in real time.
作者
刘修颀
徐宏宇
LIU Xiuqi;XU Hongyu(School of Electronic Information Engineering,Shenyang Aerospace University,Shenyang 110136,China)
出处
《现代信息科技》
2023年第12期85-88,92,共5页
Modern Information Technology
