Vehicle state estimation for INS/GPS aided by sensors fusion and SCKF-based algorithm. (March 2021)

• Record Type: Journal Article
• Authors: Song, Rui; Fang, Yongchun
• Is Part Of: Mechanical systems and signal processing. Volume 150(2021)
• Publication Date: 2021-03
• Subjects: Vehicle dynamics -- State estimation -- Robust observer -- Integrated navigation; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2020.107315
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A robust SMO is introduced to improve the capability when uncertainty are involved in the vehicle dynamic model. The coupling effect between sensors drift and vehicle motion is studied, and the measurement error is compensated by the modified ENN. The extended SCKF employs measurement from different sensors by a fusion strategy to deal with the severe driving motions. Simulation and field tests with different maneuvers are implemented. Abstract: To improve the safety and stability of land vehicles, this paper explores the estimation problem for vehicle states, including lateral velocity and attitude. First, a robust sliding mode observer is introduced to improve the adaptability for uncertain inputs, especially for the varying parameters in the vehicle dynamic model and longitudinal velocity. Furthermore, theoretical studies are performed to enhance the capability of the observer. In order to mitigate errors with the integrated navigation system, sensor drift model is primarily established based on a modified Elman neural network, so as to investigate the coupling between driving motion and errors. In addition, an extended square-root cubature Kalman filter is proposed to combine measurements from different sensors, utilizing a fusion strategy, to deal with severe driving motion and state estimation problems. Finally, simulation and field tests are carried out under a variety of maneuvers and conditions. The approach is compared with existing methods and evaluated experimentally, which indicates its effectiveness in improving the accuracy of vehicle state estimation.