Improved PCA Based Step Direction Estimation for Dead-Reckoning Localization

Step direction estimation is one of the key procedures for step counting based dead-reckoning tracking using motion sensors. It is also quite challenging, especially when the captured motion data is tainted by the user´s activity. The Principal Component Analysis (PCA) based algorithm has provided robust estimation results, regardless of the sensor´s relative rotation compared to the human body. However, the PCA based algorithm only returns the principal axis, resolving the 180-degree ambiguity is another challenge. In this paper, the drawback of PCA is compensated with the sensor´s orientation analysis, which returns the walking direction by analyzing the change of the sensor´s orientation. In our adaptive method, the sensor´s orientation analysis algorithm is executed when a direction change is detected by the PCA algorithm. Because of the low computational complexity and restricted usage of orientation analysis, the adaptive method introduces little overhead compared to the original PCA method. Experimental results show that the adaptive algorithm provides more robust and accurate results compared to the PCA algorithm.