改进的DGPS/INS/视觉组合导航算法研究  

Research on Improved DGPS/INS/Vision Integrated Navigation Technique

作　　者：姬张建[1,2] 袁运斌[1] 柴艳菊[1] 盛传贞[1,2] 马洋[1,2] 

机构地区：[1]中国科学院测量与地球物理研究所,武汉430077 [2]中国科学院研究生院,北京100049

出　　处：《系统仿真学报》2011年第12期2738-2743,2749,共7页Journal of System Simulation

基　　金：国家杰出青年科学基金(40625013);国家863计划项目(2007AA12Z311);国家自然科学基金(4089016040874018)

摘　　要：为了满足智能车辆的高精度、实时性和高可靠性的自主导航的需要,提出了一种GPS/机器视觉共同辅助SINS的多采样的智能车辆组合导航算法。该算法不仅包括GPS和SINS形成的位姿速度观测信息,还包括机器视觉形成的位置观测信息。在子滤波器中采用自适应选权滤波算法,抑制了滤波发散,提高了滤波精度。还提出了一种改进的基于估计协方差阵的奇异值分解的动态自适应调节信息分配系数的算法,可有效提高系统的状态估计精度。通过仿真实验验证,该导航系统能为智能车辆提供丰富的导航信息,实现了厘米级的导航精度和容错性,即使在GPS出现较长时间的中断时,仍能为智能车辆提供可靠的导航信息。To satisfy the autonomous navigation demands such as good accuracy,real time and high reliability for intelligent vehicle,a new multi-sample integrated navigation method,in which SINS was aided by GPS/Vision sensors,was proposed.In the method,the observation information includes two parts： one is the position velocity and attitude informed by SINS/GPS while the other is the position informed by Vision.The adaptive filtering by selecting the parameter weights was adopted in the sub-filter,which braked divergence of filter and enhances precision of filter.A new method was proposed which determined the information-sharing factors dynamically and adaptively based on singular value decomposition of the covariance matrix of the estimated errors.The method could effectively improve the precision of the estimated errors.Simulation results show that the proposed integrated navigation system can provide abundant navigation information with cm-level navigation accuracy and good fault-tolerant performance.Even when GPS is continuously interrupted for a period,the intelligent vehicle can still obtain reliable navigation information by the integrated navigation method.

关 键 词：组合导航 自适应滤波 奇异值分解 联邦滤波 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]