Dead reckoning on smartphones to reduce GPS usage

摘要

GPS localization drains battery aggressively if it is done continuously on a smartphone. This high battery drainage could be reduced by lowering the sampling rate of GPS. However, to fill the reception gaps introduced by low sampling rate, other methods need to be explored. The sensors available in the smartphones allow incremental positioning based on dead reckoning. Even though dead reckoning is a widely researched topic in navigation and walking based localization, it still lacks research on smartphone low-quality sensor based localization for vehicular motion. Therefore this paper describes a dead reckoning system design and implementation for vehicular motion on Android smartphones. It adapts and presents currently used methods for each system step (i.e. filtering, gravity removal, heading estimation and velocity/displacement integration). It also presents new research finding on accelerometer filtering process (i.e. the relationship between acceleration/deceleration to main frequency components of the motion).