张兴盛,邢科新.基于激光与视觉数据融合的改进SLAM 算法[J].高技术通讯(中文),2023,33(3):305~313 |
基于激光与视觉数据融合的改进SLAM 算法 |
Improved SLAM algorithm based on laser and vision data fusion |
|
DOI:10. 3772/ j. issn. 1002-0470. 2023. 03. 009 |
中文关键词: 激光;视觉;同时定位与建图(SLAM);数据融合 |
英文关键词: laser, visual, simultaneous localization and mapping(SLAM), data fusion |
基金项目: |
作者 | 单位 | 张兴盛 | (浙江工业大学信息工程学院 杭州310023) | 邢科新 | (浙江工业大学信息工程学院 杭州310023) |
|
摘要点击次数: 1053 |
全文下载次数: 1493 |
中文摘要: |
为了提高建图的鲁棒性,本文提出了一种基于激光同时定位与建图( SLAM) 与视觉SLAM 融合的新算法。传统算法通过传感器数据的距离信息变化量来估计其姿态,在大多数情况下能准确定位机器人的姿态。然而在退化环境中,例如机器人在长廊或者沿着单面的墙壁运动时,传感器周围环境结构特征基本无变化。此时,来自激光传感器的数据不随时间和运动而变化。针对这一问题,本文提出了一种基于激光与视觉数据融合的改进SLAM 算法。其主要思想是根据激光测量的结果引入角度置信度,对2 种传感器的数据进行加权融合。2 种传感器的权值将随角度的变化而变化。实验结果表明,与传统激光SLAM 算法相比,算法融合视觉数据后,在走廊等结构退化环境中能实现更好的定位效果,同时建图结果优于传统激光SLAM 算法。 |
英文摘要: |
In order to achieve robust mapping,this paper proposes a new method based on the fusion of laser simultaneous localization and mapping (SLAM) and visual SLAM. The traditional laser algorithm estimates the pose by the change of distance information of sensor data. This method can accurately locate the pose of the robot in most cases.However, in the degraded environment, for example, when the robot moves in a corridor or along a single wall, the structural characteristics of the surrounding environment of the sensor are basically unchanged. At this time, the data from the laser sensor does not change with time and motion. In order to solve this problem, this paper proposes an improved SLAM method based on laser and vision data fusion. The main idea is to fuse the data of the two kinds of sensors by introducing angle confidence according to the results of laser measurement. The experimental results show that, compared with the traditional laser SLAM algorithm, the algorithm can achieve a better positioning effect in degeneration environment and its mapping results are better than the traditional laser SLAM algorithm. |
查看全文
查看/发表评论 下载PDF阅读器 |
关闭 |
|
|
|