Research on Accuracy Analysis and Enhancement of Low-cost MEMS INS/GNSS Integration for Land Vehicle Navigation
Quan Zhang and Xiaoji Niu, Wuhan University, China
Location: Big Sur
With the development of application requirements, land vehicle navigation requires high precision (less than meter-level positioning accuracy) and high reliability to satisfy the needs of all road conditions including the scenario where GNSS information is unavailable for a short time (e.g., one minutes) because of signal blockage. INS/GNSS integrated system can provide accurate and reliable navigation information (including position, velocity and attitude), and significantly improve positioning continuity compared to the individual INS or GNSS. However, situations of complete GNSS gap may occur in urban environments or in the scenarios of signal blockage, the INS navigation solution drifts overtime due to its inherent sensor errors. Recently, MEMS IMU has become commercially available and been widely applied in possible applications such as land vehicle navigation at low cost. Low-cost MEMS INS cannot work independently for a long time and the navigation solution get quick divergent with time due to the poor inertial sensor performance. In such cases, in addition to GNSS information assistance, other auxiliary information, especially the motion constraints to reduce the need for external sensors, is required to ensure the navigation performance. Therefore, the paper mainly focuses on the research of performance analysis of low-cost MEMS INS/GNSS integration, and further proposes effective accuracy enhancement methods using motion constraints to meet the application requirements.
The paper works on accuracy analysis and enhancement of low-cost MEMS INS/GNSS integration for land vehicle navigation. Firstly, a performance comparative analysis of the different grade MEMS IMUs (high-, medium- and low-cost) from Sensonor, ADIS and Murata is presented. GNSS outages are simulated in clean datasets collected in open sky conditions, by artificially omitting satellites during data processing to assess the stand-alone navigation performance of each MEMS IMU. Secondly, in the paper, several different constrained motion attributes are treated as additional fictitious or pseudo measurements to aid inertial navigation solution and prevent INS error accumulation in the GNSS gap. Constrained motion attributes such as Non-holonomic constrain (NHC) of land vehicle, which assume that some navigation parameters remain almost constant for short time period based on the kinematic constraints, can be usually applied to suppress the drift of INS solution. Except the well-known NHC, the paper applies the height constraint and angular rate constraint based on the motion attributes of land vehicle to enhance the auxiliary effect. Thirdly, the paper proposes to take the kinematic constrain as state constraints with equality and/or inequality incorporated in the optimal estimation algorithm to design constrained Kalman filter to enhance navigation accuracy and reliability. Finally, the datasets collected in open sky conditions (simulating GNSS outages) and typical urban environment are both used to analyze the effectiveness of proposed enhancement method. The results show that constrained motion attributes can suppress the navigation drift errors in GNSS outages, and improve the navigation reliability.
In summary, this paper mainly analyzes the several kinematic constraints to aid INS solution, and proposes to apply the Kalman filter with state equality/inequality constraints using motion attributes of land vehicle to enhance GNSS/INS integrated system to provide navigation solution of high accuracy and high reliability.