摘要
针对低成本IMU自主个人导航系统方位漂移的问题,提出了一种融合鞋式IMU、楼层平面图的个人导航解决方案,为实现精度较高的室内相对定位设计了一种新的辅助粒子滤波算法。引入卡尔曼滤波+粒子滤波的级联框架,底层卡尔曼滤波器在捷联解算的基础上,利用零速修正技术估计IMU的位置和姿态;上层粒子滤波器提取步行中每一步的步长和方位变化作为量测,建立相应的步行运动模型融入非线性地图匹配技术。考虑室内应用情境,通过对传递粒子的多步推演预测和选择性剔除,推导了一种新的粒子滤波算法。采集低精度IMU的室内行走数据验证了算法的有效性:约300 m行程的室内导航最终位置误差不超过0.3 m。表明提出的级联滤波算法框架能有效解决建筑平面信息辅助情形下的个人导航问题,新设计的粒子滤波算法有助于提高个人导航系统连续位置测定的精度和可靠性。
As heading drift error remains a problem in a standalone pedestrian navigation system（PNS） that uses low-cost inertial measurement unit（IMU）,an algorithm for integrating shoe-mounted IMU with building plane was proposed,and a novel auxiliary particle filter which is more applicable in indoor navigation scenario was devised.A double-deck framework comprises Kalman filter（KF） and particle filter（PF） was introduced,in which the lower KF applies zero-updating measurement for drift correction.The upper PF computes the step-wise changes of IMU position and heading to use them as measurements,and a corresponding pedestrian movement model was constructed for fusing nonlinear map-matching technique.The proposed algorithm is verified through experimental data collected from a low-performance IMU mounted on foot：the final positioning error of 300 m travel distance is less than 0.3 m.It is also shown that the consistent positioning accuracy and reliability of a PNS could be improved effectively with the modified auxiliary particle filter.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2013年第1期1-6,共6页
Journal of Chinese Inertial Technology
基金
国家自然科学基金资助(61273333)
总装备部惯性技术预研基金资助(51309040501)
关键词
个人导航
室内定位
辅助粒子滤波
地图匹配
pedestrian navigation
indoor positioning
auxiliary particle filter
map-matching
