Vector graph assisted pedestrian dead reckoning using an unconstrained smartphone.

摘要: The paper presents a hybrid indoor positioning solution based on pedestrian dead reckoning (PDR) approach using built-in sensors smartphone. To address the challenges of flexible and complex contexts carrying phone while walking, robust step detection algorithm motion-awareness has been proposed. Given fact that length is influenced by different motion states, an adaptive estimation recognition developed. Heading carried out attitude acquisition algorithm, which contains two-phase filter to mitigate distortion magnetic anomalies. In order estimate heading for unconstrained smartphone, principal component analysis (PCA) acceleration applied determine offset between orientation smartphone actual pedestrian. Moreover, particle with vector graph assisted weighting introduced correct deviation in estimation. Extensive field tests, including four phone, have conducted office building verify performance proposed algorithm. Test results show can achieve sub-meter mean error all contexts.