Register    Attendee Sign In Sign in to access papers, presentations, and videos
Previous Abstract Return to Session E4b

Session E4b: Smartphone Decimeter Challenge (Co-sponsored by Google)

First Place Winner of the Smartphone Decimeter Challenge: Optimized GNSS/INS Integration Approach for Smartphone Positioning
Norizumi Motooka, Mitsubishi Electric Corporation
Date/Time: Thursday, Sep. 19, 5:08 p.m.

Peer Reviewed

This study presents a tightly coupled integration method of the global navigation satellite system (GNSS) and inertial navigation system (INS) to accurately determine locations using smartphone data. Data utilized in the Google smartphone decimeter challenge (GSDC) 2023-2024 involved participants tasked with determining the location of a smartphone mounted on the dashboard of a moving vehicle. In previous GSDC events (2021, 2022), few studies focused on positioning methods using inertial measurement unit (IMU) data. This was primarily because of the challenge of synchronizing IMU data timestamps with GPS time as the former is driven by the system clock. However, GNSS/INS integration, a widely used technique for ground vehicle navigation, proved to be a suitable approach for achieving high-precision positions in the provided dataset. The stability of the IMU's mounting in a moving vehicle, coupled with data collection in open-sky environments where signal interference is minimal, made GNSS/INS integration an effective method. To facilitate GNSS/INS integration, this study proposes a method to adjust IMU data timestamps to ensure consistency with GNSS observations. Additionally, an adaptive measurement weighting scheme is introduced to address the varying quality of GNSS measurements obtained from different driving environments and various smartphone models. The implementation of these methods significantly improved the positioning accuracy, with enhancements of 36 cm and 14 cm, respectively, in the score, the mean of the 50th and 95th percentile distance errors. The final public and private scores achieved were 0.905 m and 0.883 m, respectively, marking a milestone achievement of accuracy within 1 m and securing the first position at GSDC 2023-2024.



Previous Abstract Return to Session E4b