基于多状态约束的惯性/里程计信息融合方法研究

Highly-accurate INS/odometer autonomous land navigation system is vital to both military and many civilian applications. As a relative measurement sensor, the odometer measures the trajectory length that is dependent on vehicle’s all motion states over the time interval. As traditional estimators can only handle instantaneous observation, the main-stream research uses an odometer as a velocity sensor, namely, deriving velocity from the raw trajectory length measurement, which is unable to fully exploit the property of the odometer measurement and its effectiveness in adding INS. Recent advances and successful applications in the related fields of underwater acoustic positioning and visual motion estimation indicate that the proper processing of relative-measurement sensor information can significantly improve the system performance, which opens a new door for INS/odometer research. With the latest achievements in relative-measurement sensor information fusion, this project will investigate the multi-state constraint based INS/odometer information fusion techniques and make full use of attributes of the odometer’s trajectory length measurement, aiming to elevate our autonomous land navigation system’s performance and accuracy level and providing solid support for military precision orientation determination and positioning and civilian car navigation. In the meantime, through this specific INS/odometer research, this project will foster general theory, techniques and methods for general inertial-based navigation systems.

高精度惯性/里程计自主导航系统对军民车辆应用至关重要。作为一种相对测量传感器，里程计所测量的行程信息与对应时间段内的所有运动状态都有关。因传统状态估计器只能处理瞬时观测，国内外研究大都将里程计用作测速仪，即利用由里程计行程测量导出的速度信息，未能充分发掘里程计的行程测量特点及其对惯性导航系统的辅助能力。相关领域特别是水下声学定位和视觉运动估计的最新进展及成功应用表明：相对测量传感器信息的妥善处理可显著提升系统精度，为惯性/里程计导航研究开启了一扇崭新的大门。本项目拟借鉴相对测量传感器信息融合的最新成果，研究基于多状态约束的惯性/里程计信息融合技术，充分发掘和利用里程计行程信息的特点，进一步提升我国陆用自主导航系统的性能和精度水平，为军用精密定位定向和民用车辆导航应用提供有力支持。同时，通过惯性/里程计这一特定研究对象，凝练适用于一般惯性基组合导航系统的理论、技术和方法。

惯性/里程计导航是导航领域的经典问题，高精度惯性/里程计自主导航系统对车辆应用至关重要。本项目系统研究了惯性/里程计信息融合技术，充分发掘了里程计行程信息的特点，分析了里程计测量原理以及当前测量方法的误差来源，研究了里程计累积脉冲测量、脉冲增量测量和脉冲速度测量的误差特性，提出使用多模型自适应滤波算法考虑测量误差特性的变化。长距离实际系统测试表明，导航定位精度优于0.2‰D（导航级IMU）。通过惯性/里程计这一特定研究对象，凝练出鼎立四元数惯导描述、精确惯导解算等适用于一般惯性基组合导航系统的理论、技术和方法。.项目在惯性/里程计组合导航、惯导描述与精确解算、低成本惯导标定与步行导航应用等方面做出了引领性工作，核心成果在领域顶级会议上报告、并发表在领域顶刊上，受邀报告12次，发表IEEE汇刊论文8篇，申请专利8项（授权4项）。项目负责人连续入选Elsevier中国高被引学者，受邀担任IEEE AESS杰出讲师、德国导航协会会议DGON ISS常任程序委员，俄罗斯导航协会会议ICINS程序委员。