A Quasi-Monte Carlo Based Gaussian Particle Filter

摘要：

A novel Gaussian particle filter based on Quasi-Monte Carlo integration is proposed for nonlinear filtering problems with Gaussian distributions.The conventional Gaussian particle filter,with Iess complexity than particle filter,also has the same problem that approximation accuracy deteriorating caused by the randomness of particles from Monte-Carlo sampling.The proposed algorithm,called QMC-GPF,circumvents this difficulty by placing the particles deterministically according to a Quasi-Monte Carlo integration rule.With weighted low-discrepancy particles in place of weighted random samples,the Quasi-Monte Carlo methods can obtain better approximation performance in nonlinear estimation.Experimental results with a simulated 2D tracking problem demonstrates the effectiveness of the proposed new QMC-based filtering algorithm.

作者: Bin Wu Hong-Bing JI Xi CHEN

会议类型: 国际会议

会议名称: The International Conference Information Computing and Automation(2007国际信息计算与自动化会议)

会议地点: 成都

会议语种:英文

页码: 285-288

在线出版日期: 2007-12-19（万方平台首次上网日期，不代表论文的发表时间）