Multi-sensor Fusion Based GNSS Differential Enhancement Research

As an all-weather and high-precision system,the Global Navigation Satellite System(GNSS)plays an indispensable role in the national production and people's life for navigation,positioning,and timing.Currently,GNSS positioning is restricted by two scenarios.Firstly,the noise in smartphones with low-cost antenna and chipset impairs positioning accuracy greatly;Secondly,GNSS signals cannot pass through buildings,making it difficult to perform well as the opening fields.In contrast,IMU and vision-based localization technology can enhance GNSS positioning.In this paper we focus on the GNSS enhancement algorithm on smartphones and in the scenes of weak GNSS signals.Firstly,we implement the double-difference algorithm of pseudorange and Doppler in smartphones and analyzes the property of raw data of GNSS chipsets.Further,a fusion algorithm with IMU is developed to increase positioning accuracy.To make continuous localization in the weak GNSS signal scenes,a multi-sensor fusion framework is developed.The framework is based on the error-state Kalman Filter and supports the modular addition of sensors by separating the main state and the auxiliary state.The process of multi-reference frame calibration and initialization is also analyzed.The experiments show that the differential algorithm effectively improves the positioning accuracy on smartphones.And with the multi-sensor fusion framework,one can achieve continuous positioning indoor and outdoor at the condition of a wide range of complex scenes for a long time.