论文部分内容阅读
传统基于微机电惯性测量单元(MEMS-IMU)的惯性导航系统(INS)引入零速修正(ZUPT)算法校正器件的累积误差。但由于ZUPT零速判定阈值为固定值,只适合单一运动模式,当室内行人运动轨迹包含多种运动模式时,定位精度下降。对此,提出了一种多运动模式下自适应阈值零速修正算法。分析了室内行人包括静止(Still)、走(Walk)、跑(Run)、上楼(Upstairs)和下楼(Downstairs)五种运动模式零速判定阈值的选取,实现了利用随机森林(Random Forest,RF)算法对五种运动模式分类识别,并根据识别结果对ZUPT零速判定阈值进行自适应调整。为了验证算法的可行性和有效性,利用MATLAB软件平台对实测数据进行处理,并与传统定位算法进行了比较。三组实验结果表明,当室内行人运动轨迹包括多种运动模式时,相比传统固定阈值的ZUPT算法,引入自适应调整阈值的ZUPT算法可使定位算法的定位精度提高73.83%。
Conventional inertial navigation systems (INS) based on MEMS-IMUs introduce the zero-speed correction (ZUPT) algorithm to correct the cumulative error of the device. However, since the ZUPT zero-speed determination threshold is a fixed value, it is only suitable for a single motion mode. When the indoor motion trajectory includes multiple motion modes, the positioning accuracy decreases. In this regard, an adaptive threshold zero-speed correction algorithm in multi-sport mode is proposed. This paper analyzes the selection of zero-speed thresholds for indoor pedestrians such as Still, Walk, Upstairs and Downstairs, and makes use of Random Forest , RF) algorithm to classify the five kinds of motion patterns and adaptively adjust ZUPT zero-speed decision threshold according to the recognition results. In order to verify the feasibility and effectiveness of the algorithm, the measured data is processed by MATLAB software platform and compared with the traditional positioning algorithm. The experimental results of three groups show that ZUPT algorithm with adaptive threshold adjustment can improve the positioning accuracy of the localization algorithm by 73.83% compared with the traditional fixed threshold ZUPT algorithm when the indoor pedestrian motion trajectories include multiple motion modes.