摘要
基于足绑式惯性测量单元(IMU)的惯性导航系统被广泛应用于行人导航中,其通过零速修正(ZUPT)算法可对速度估计误差进行较好的补偿,然而其位置误差会随时间发散。针对于此,提出了一种基于室内合作场景智能识别的行人导航算法。通过随机森林算法,对行人在室内平地步行、上楼梯、下楼梯等不同步态进行训练与辨识,并结合室内先验地图对行人导航的结果进行校正。通过实验表明,行人在室内行走1100m时最大定位误差为1.85m(总行程0.17%),相对无场景识别的方法精度提高了6倍,可以有效提高行人导航精度。
Inertial navigation system based on foot-mounted IMU(Inertial Measurement Unit) is widely used in pedestrian navigation, and it can restrain the velocity error by the zero velocity update(ZUPT) algorithm, but the position error will diverge with time. Aiming at this, a pedestrian navigation algorithm based on intelligent recognition of indoor cooperative scene is proposed. The random forest algorithm is used to train and identify the pedestrian’s gait in indoor scene, such as walking on the ground, going up the stairs, going down the stairs, and the pedestrian’s navigation result is corrected by combining the priori maps of buildings. The experimental result shows that the maximum positioning error is 1.85 m(0.17% of the total distance) when the pedestrian walks 1100 m in the building, which is 6 times more accurate than the method without scene recognition and able to effectively improve the pedestrian navigation accuracy.
作者
朱超群
赖际舟
吕品
叶素芬
袁诚
ZHU Chao-qun;LAI Ji-zhou;LYU Pin;YE Su-fen;YUAN Cheng(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;Jiangsu University Key Laboratory of Internet of Things and Control Technology,Nanjing 211106,China)
出处
《导航定位与授时》
2019年第6期63-69,共7页
Navigation Positioning and Timing
基金
国家自然科学基金(61703207)
江苏省自然科学基金(BK20170801)
航空科学基金(2017ZC52017)
中央高校基本科研业务费专项资金(NG2019001,NT2019008)
关键词
行人导航
惯性测量单元
零速修正
场景识别
Pedestrian navigation
Inertial measurement unit
Zero velocity update
Scene recognition
