Sit-Down & Stand-Up Awareness Algorithm for the Pedestrian Dead Reckoning

Abstract

In this paper we present an context awareness algorithm of sit-down, stand-up and walks behaviors of a pedestrian for an adaptive step length estimation algorithm of PDR (Pedestrian Dead Reckoning). PDR consists of MEMS accelerometers, gyros and magnetometers. In general PDR is attached to a body of the pedestrian and provides the position information of the pedestrian. The developed PDR is assumed that it is embedded in the cellular phone. Thus only a single sensor module is used to develop the context awareness algorithm. In order to compute the walking distance precisely, various behaviors must be classified estimated as well. PDR detects steps of the pedestrian and estimate step length and heading. It means that large step detection error can cause large position error in PDR. Therefore, it is necessary for PDR the context awareness algorithm. In this paper, we classified sit-down, stand-up and walking of pedestrian using a variance of horizontal attitudes (roll and pitch) of the PDR module and the norm of the horizontal accelerations in the navigation frame. The developed method can be applied to USN (Ubiquitous Sensor Network) and u-health monitoring system for the mobile system.