摘要
针对广泛应用于定位与导航、目标跟踪和模式识别的粒子滤波算法,通过仿真实验研究了不同的估计输出、采样阈值、重采样算法以及运动模型和观测模型不准确对粒子滤波效果的影响,仿真实验结果表明,粒子滤波器鲁棒性较好。
Particle filter can both solve linear,Gaussian problems and nonlinear,non-Gaussian ones.Research work is carried out on the particle filter algorithm which has been widely used in location and navigation,object tracking and pattern recognition.Firstly,the basic algorithm of the particle filter is described.Secondly,through the simulation experiments,a research is made on the effect of the estimated output,sampling threshold,resampling algorithm and the error motion model or error observation model on the filter.Finally,through the experiments analysis,it is concluded that the particle filter has good robustness.
出处
《北京信息科技大学学报（自然科学版）》
2011年第2期82-87,共6页
Journal of Beijing Information Science and Technology University
基金
北京市属高等学校人才强教深化计划资助项目(PHR201106131)
关键词
粒子滤波器
重采样
估计输出
采样阈值
运动模型
观测模型
particle filter
resampling
estimated output
sampling threshold
motion model
observation model
