基于LightGBM的AGV多传感器数据融合方法
首发时间:2020-03-19
摘要:针对AGV自主导向车的定位问题,在研究大量文献的基础上,提出一种基于LightGBM的AGV多传感器数据融合方法。使用LightGBM模型对传感器定位数据进行融合,将每组传感器模型得到的坐标数据视为一类特征,针对两个方向上的坐标数据分别训练两个LightGBM模型,利用多组特征数据组成的数据集对模型进行训练调参,通过组合多组传感器的测量数据来降低不确定性,从而获得高精度的定位信息。最后,通过真实场景实验与基于卡尔曼滤波的融合算法比较,验证了本方法对AGV定位精度有更加明显的提升,定位精度可达6.3 mm。
关键词: 传感器 多传感器数据融合 AGV定位 LightGBM模型
For information in English, please click here
A multi-sensor data fusion method based on LightGBM model for AGV localization
Abstract:Aiming at the positioning problem of automated guided vehicles, based on a large amount of literature research, a LightGBM-based AGV multi-sensor data fusion method was proposed. The LightGBM model is used to fuse the sensor positioning data, and the coordinate data obtained from each group of sensor models is regarded as a type of feature. Models are trained for two LightGBM coordinate data in both directions, using a plurality of sets of data consisting of characteristic data set to train the model parameter adjustment. A plurality of sets of measured data by the sensor was combined to reduce the uncertainty, thereby obtaining a highly accurate positioning information.Finally, through real-world experiments comparing with the Kalman filter fusion algorithm, the method has verified more significantly improved positioning accuracy of the AGV, the positioning accuracy of up to 6.3 mm.
Keywords: sensor multi-sensor data fusion AGV positioning LightGBM model
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于LightGBM的AGV多传感器数据融合方法
评论
全部评论0/1000