A Novel Three Dimensional Movement Model for Pedestrian Navigation

Abstract

In this paper, a Three Dimensional Pedestrian Movement Model (3D-MM) capable of probabilistically representing pedestrian movement in challenging indoor and outdoor localization environments is developed, implemented and evaluated. In the scope of this paper, the model is used to generate a �movement� or a transition for dynamic positioning systems that are based on sequential Bayesian filtering techniques, such as particle filtering. It can also be used to assign weights for particles' movements proposed by sensors in Likelihood Particle Filters implementations. Alternatively, the developed model can be applied to other applications domains such as infrastructure design, evacuation planning, robot-human interaction and pervasive computing. The novelty of the model is in its ability to characterize both random and goals-oriented pedestrian motions and additionally use the a priori knowledge of maps and floor plans. It will be shown that an appropriate pedestrian movement model not only improves the positioning accuracy, but is also essential for a robust positioning estimator. Additionally, this work shows that maps and floor plans can improve pedestrian movement models but do not replace them, as several authors suggest.