Skip to main content
Open Access Publications from the University of California

UC Riverside

UC Riverside Electronic Theses and Dissertations bannerUC Riverside

Real-Time Sub-Meter Vehicle Positioning: Low-Cost GNSS-Aided INS

  • Author(s): Rahman, Farzana S.
  • Advisor(s): Farrell, Jay A
  • et al.

Many applications, including connected and autonomous vehicles, would benefit from navigation technologies that reliably achieve sub-meter position accuracy. Real-time submeter Earth-referenced positioning accuracy has the potential to be achieved with high reliability on moving vehicles by using Global Navigation Satellite Systems (GNSS) if common-mode ranging error correction information is communicated to the vehicles. For successful commercial implementation, such correction information must be delivered on continental or global scales. The communication latency must be small enough to not significantly affect performance. The main focus of the study is to investigate the feasibility to achieve sub-meter positioning accuracy with low-cost GNSS receiver and Inertial Measurement Unit (IMU) sensor. The thesis is divided into two phases based on their common-mode compensation approach. The first phase considers Local Area Differential GNSS (LADGNSS) approach and the second phase investigates Wide Area Differential GNSS (WADGNSS) method.

The first part of this project presents a local differential correction computation methodology designed to be robust to latency and studied position estimation accuracy as a function of differential correction latency for both stationary and moving receivers [121, 122]. This performance was robust to latency up to 600 seconds. Two algorithms incorporated the local differential correction approach defined in [122]. The Position, Velocity, Acceleration (PVA) approach used only DGNSS data with a Kalman filter. The Inertial Navigation System (INS) approach used DGNSS and inertial measurement data within an extended Kalman filter(EKF). The study showed that both approaches achieved performance exceeding the Society of Automotive Engineering (SAE) J2945 specification (1.5 meter horizontal accuracy and 3.0 meter vertical accuracy at 68%) [12] with PVA achieving 1 m horizontal accuracy at 90% and 2 m vertical accuracy at 95% and the INS approach using a consumer-grade IMU achieved 1 m horizontal accuracy at 98% and 2 m vertical accuracy at 95% [125].

The second phase of this project investigates methods for implementing DGNSS corrections on a continental scale, to study the achievable accuracy. The overview includes discussion of WADGNSS, the models that it incorporates, the modeling agencies, and the existing data and model sources. The paper presents a PPP aided INS design and analyzes navigation performance as a function of IMU quality. This paper considers GNSS Precise Point Positioning (PPP) with Least Square (LS), PVA and EKF for static and only EKF for dynamic condition. The experimental results demonstrate positioning accuracy that surpasses the SAE specification using PPP corrected single frequency, single constellation GNSS measurements along with a consumer-grade IMU sensor. Experiments performed in this project (see Section 5.3.2) have demonstrated horizontal position accuracies of 1.35 0.48, 1.19 0.41, and 0.47 0.26 using PPP PVA and demonstrated horizontal position accuracies of 0.81 0.21, 0.52 0.25, and 0.43 0.186 using PPP INS for stationary dataset. Horizontal and vertical position accuracies (see Section 5.3.3) are 0.80 0.40 and 2.32 1.14 with PPP INS respectively for dynamic condition.

This study focuses on single frequency, single constellation results. The availability of multiple constellations and multiple frequencies per constellation will facilitate estimation and compensation of ionospheric error, accommodation of outliers, and accommodation of multipath. It will also greatly increase the number of available measurements and the likelihood that the user has available

a set of satellites with appropriate geometry to reliably achieve the performance specification.

Main Content
Current View