自适应序贯滤波的机器人融合定位方法研究
Robot Fusion Localization Method Based on Adaptive Sequential Filtering
孙凌宇 1李庆翔 1李鑫宝 1王子航1
作者信息
- 1. 河北工业大学机械工程学院,天津 300401
- 折叠
摘要
针对激光雷达在长直环境下鲁棒性低以及视觉相机受光照条件影响大的问题,提出一种利用序贯卡尔曼滤波将两种传感器采集到的信息进行融合的定位方法,将传感器作为独立的节点逐级传递以实现多层次滤波,解决了异步传感器更新延迟导致算法中断的问题,同时为算法添加自适应成分,避免了移动机器人长距离行驶过程中因缺少动态调节而产生发散的现象.仿真实验结果表明,以自适应序贯滤波的形式,融合定位结果显著降低了单一传感器测量带来的误差,有效提升了移动机器人在未知环境下的定位精度.
Abstract
Aiming at the low robustness of lidar in long straight environment and the problem that the visual camera is greatly af-fected by the illumination conditions,this paper proposes a positioning method using sequential Kalman filter to fuse the informa-tion collected by the two sensors,which transfers the sensors as independent nodes step by step to realize multi-level filtering.At the same time,the adaptive component is added to the algorithm to avoid the divergence phenomenon caused by the lack of dynamic adjustment in the long-distance driving process of mobile robot.Simulation results show that in the form of adaptive sequential fil-tering,the fusion positioning results significantly reduce the error caused by single sensor measurement,and effectively improve the positioning accuracy of mobile robot in unknown environment.
关键词
序贯滤波/多传感融合/机器人技术/激光雷达/视觉相机Key words
Sequential Filtering/Multi Sensor Fusion/Robotics/Lidar/Visual Camera引用本文复制引用
基金项目
国家自然科学基金委员会自主项目(U1913211)
出版年
2024
NETL
NSTL
万方数据