里程表
全球导航卫星系统应用
气压计
全球定位系统
计算机科学
遥感
全球导航卫星系统增强
实时计算
电信
地理
人工智能
气象学
作者
Kai‐Wei Chiang,Hao‐Wei Chang,Yuhua Li,Guang-Je Tsai,Chung-Lin Tseng,Yu-Chi Tien,Pei-Ching Hsu
出处
期刊:IEEE Sensors Journal
[Institute of Electrical and Electronics Engineers]
日期:2020-03-15
卷期号:20 (6): 3057-3069
被引量:39
标识
DOI:10.1109/jsen.2019.2954532
摘要
With the increasing demands for seamless landvehicle navigation, systems with robust performance are required in highly urbanized areas. The traditional INS/GNSS integration is widely applied to solve this issue. However, the system still suffers from bad GNSS signal reception and INS time accumulated errors that seamlessness and stability are difficult to maintain. In this study, the performance of the low cost INS/GNSS with aiding sensors, such as an odometer and barometer, was evaluated for both the loosely-coupled (LC) scheme and tightly-coupled (TC) scheme. Moreover, considering barometric error accumulation, a vehicle-behavior based drift control method has been proposed. An experiment was conducted under harsh GNSS-degraded scenarios to assess the characteristics and performance for different sensor combinations, using single constellation (GPS) with single-frequency (L1 band) measurement. Overall, the TC scheme without additional strategies in detecting abnormal measurement encounters more challenges to achieve stable performance. In an INS/GNSS/barometer system with the proposed drift control method, error accumulation under unpredictable environmental changes was successfully mitigated in both schemes. The proposed method can maintain a height accuracy of 2-meter level root mean square even after a long term operation. In an INS/GNSS/odometer combination, improvements were observed in the horizontal and vertical direction for both schemes. According to statistical analysis, an INS/GNSS/odometer/barometercombination shows 16.57% and 6.11% in the horizontal, and 30.71% and 71.28% in the vertical for the LC scheme and TC scheme, compared with an INS/GNSS combination.
科研通智能强力驱动
Strongly Powered by AbleSci AI