流形上迭代扩展卡尔曼滤波器快速指南
A Quick Guide for the Iterated Extended Kalman Filter on Manifolds
AI总结:
针对现有流形上迭代扩展卡尔曼滤波器(IEKF)相关阐述冗长艰深且易错的问题,本文以大一水平矩阵微积分为基础,提供了含核心方程及推导关键步骤的流形上IEKF快速参考指南。
摘要:
扩展卡尔曼滤波器(EKF)是离散非线性系统常用的状态估计方法。它随时间推移递归执行传播步骤,在一组测量值到达时执行更新步骤。在更新步骤中,EKF仅对测量函数进行一次线性化。相比之下,迭代EKF(IEKF)通过迭代求解最小二乘问题在更新步骤中优化状态。IEKF已被扩展应用于状态变量位于流形上的场景,这类流形具备可微的$\boxplus$和$\boxminus$算子,李群也包含在内。然而现有相关阐述往往篇幅冗长、内容艰深,甚至存在错误。本说明以大一水平的矩阵微积分为基础,提供流形上IEKF的快速参考资料。除核心方程外,我们还重点说明了推导这些方程的关键步骤。
英文摘要:
The extended Kalman filter (EKF) is a common state estimation method for discrete nonlinear systems. It recursively executes the propagation step as time goes by and the update step when a set of measurements arrives. In the update step, the EKF linearizes the measurement function only once. In contrast, the iterated EKF (IEKF) refines the state in the update step by iteratively solving a least squares problem. The IEKF has been extended to work with state variables on manifolds which have differentiable $\boxplus$ and $\boxminus$ operators, including Lie groups. However, existing descriptions are often long, deep, and even with errors. This note provides a quick reference for the IEKF on manifolds, using freshman-level matrix calculus. Besides the bare-bone equations, we highlight the key steps in deriving them.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。