基于高阶扩维卡尔曼滤波器的SLAM算法研究
SLAM Algorithm Based on Higher-order Extended Kalman Filter
蔡彦斌 1刘继新 1邓立伟 1曹月2
作者信息
- 1. 吉林化工学院 信息与控制工程学院,吉林 132022;广东石油化工学院 自动化学院,广东 茂名 525000
- 2. 广东石油化工学院 自动化学院,广东 茂名 525000
- 折叠
摘要
针对扩展卡尔曼滤波在解决强非线性系统问题时精度不高、在机器人进行同时定位与建图时误差大、抗干扰性弱、小车运动估计效果不理想等问题,提出了一种基于高阶扩维卡尔曼滤波器的小车运动估计方法.为了解决舍入误差,高阶扩维卡尔曼滤波器算法引入隐变量,将小车运动模型进行伪线性化表示,在系统变量和隐变量之间建立线性关系,将系统状态模型转化为线性形式,再将观测模型进行等价改写.最终将小车运动模型改写为符合卡尔曼滤波的线性形式.仿真结果显示,基于高阶扩维卡尔曼滤波器的SLAM算法具有高性能和高精度等优点.
Abstract
A method based on Higher-Order Extended Kalman Filter(HEKF)is proposed to address the limitations of Extended Kalman Filter(EKF)in dealing with highly nonlinear systems,large errors in simultaneous localization and mapping(SLAM)for robots,weak disturbance rejection,and suboptimal motion estimation for vehicles.The HEKF algorithm introduces hidden variables to mitigate rounding errors,linearizes the motion model of the vehicle,establishes a linear relationship between the system variables and hidden variables,and transforms the system state model into a linear form.The observation model is also equivalently rewritten.Consequently,the motion model of the vehicle is reformulated to comply with the linear form of the Kalman filter.Simulation results demonstrate that the HEKF-SLAM algorithm exhibits superior performance and high accuracy.
关键词
SLAM/特征估计/卡尔曼滤波器Key words
SLAM/feature estimation/Kalman Filter引用本文复制引用
基金项目
广东石油化工学院人才引进项目(2020rc32)
广东省科技创新战略专项资金项目(2023S003042)
出版年
2024
国家科技期刊平台
NSTL
万方数据