基于硬件实现的粒子滤波重采样算法研究

Research of Resampling Algorithm for Particle Filter Base on Hardware Implementation

粒子滤波算法用于硬件实现是目前一个新的研究方向,传统的粒子滤波算法计算量大,所需存储空间大,实时性差,所以在硬件实现方面面临着极大的挑战。为使算法更加适合于硬件实现,以粒子滤波中的重采样步骤为研究重点,以典型的序贯重要性重采样滤波算法为例,对典型的几种重采样算法的复杂度、所需存储空间及执行时间上进行分析研究,并在TI DSPTMS320C5402上对采样算法进行仿真,结果表明部分重采样算法(PDR)更适合于硬件实现。

Recently the implementation of particle filter is a new researching field. The traditional particle filter algorithrn is excessive in computation and requires large memories. It can not meet the need of real-time processing. So, the particle filter is confi＇onted with the challenges in hardware implementation. The main researching point is the resampling of particle filter from the view of hardware implementation suitablely. Take the representative SIRF for example and compare the complexity of computation i Crequired memories and operational time： The simulation results on TI DSP TMS320VC5402 indicates that the PDR is the most suitable for hardware implementation in compared resampling algorithm.