Implementation methodology of embedded land vehicle positioning using an integrated GPS and multi sensor system

This paper presents an embedded implementation approach of land vehicle navigation involving a Multi Sensor System (MSS) consisting of a single-axis gyroscope and an odometer integrated with GPS receiver. With the assumption that the vehicle stays mostly in the horizontal plane, the vehicle speed obtained from the odometer measurements is decomposed into east and north velocities by using heading information from the gyroscope. Subsequently, the vehicle's position in latitude and longitude are determined. MSS errors are estimated by an integrated MSS/GPS Kalman filter (KF) which relies on a dynamic error model of position, velocity and heading as well as stochastic models for gyroscope and odometer errors. In case of a GPS outage, the designed KF module provides positioning information. The decentralized KF algorithm is described in software and is executed on an embedded soft core processor containing a single precision floating point unit. Results were validated imposing numerous simulated GPS outages of varied lengths on road test trajectory data of GPS receiver, car chip odometer and single axis MEMS based gyro. The length of the simulated GPS outages varied from 36 s to 425 s on three different road trajectories. Results show a maximum positional error of 110 m for an outage of 120 s duration and a minimum positional error of 14 m for an outage of 60 s duration with respect to the reference trajectory.

本文提出了一种陆地车辆导航的嵌入式实现方法，该方法涉及一个由单轴陀螺仪和与全球定位系统（GPS）接收器集成的里程表组成的多传感器系统（MSS）。假设车辆大多处于水平面内，通过使用陀螺仪的航向信息，将从里程表测量值获得的车速分解为东向和北向速度。随后，确定车辆的纬度和经度位置。多传感器系统的误差由一个集成的MSS/GPS卡尔曼滤波器（KF）进行估计，该滤波器依赖于位置、速度和航向的动态误差模型以及陀螺仪和里程表误差的随机模型。在GPS失效的情况下，所设计的KF模块提供定位信息。分散式KF算法在软件中进行描述，并在一个包含单精度浮点单元的嵌入式软核处理器上执行。通过对GPS接收器、汽车芯片里程表和基于单轴微机电系统（MEMS）的陀螺仪的道路测试轨迹数据施加大量不同时长的模拟GPS失效情况来验证结果。在三条不同的道路轨迹上，模拟GPS失效的时长从36秒到425秒不等。结果表明，相对于参考轨迹，在120秒的失效时长下，最大位置误差为110米，在60秒的失效时长下，最小位置误差为14米。