Evaluating reorientation strategies for accelerometer data from smartphones for ITS applications

Abstract

Given their ubiquity and sensing capabilities, current smartphones have been used to explore different real-life tracking and monitoring scenarios. Particularly, in the domain of Intelligent Transportation Systems (ITS), the exact orientation of the smartphone must be known to gain full advantage of the data provided by its internal accelerometers. From here, rich contextual information could be inferred. Nonetheless, in real-life scenarios, smartphones are freely placed within vehicles, so a reorientation strategy needs to be applied. The usage of several algorithms to reorient acceleration readings has been mentioned for ITS applications, but very little evaluation of their efficacy has been performed. In this work, we study the effectiveness of four algorithms for vertical reorientation, and two for triaxial reorientation of acceleration readings. Results suggest that all methods for the vertical case are equivalent, however, in the case where triaxial reorientation is needed, current strategies are far from acceptable results. We expect these findings could promote further research to alleviate these issues.