An Approach for Generating Pedestrian Network Based on Improved Navigation Graph

Abstract

Pedestrian network is the basic data for route planning in pedestrian navigation systems. The optimal route computed by the pedestrian network can be used as wayfinding cues and provide pedestrians with route guidance. Currently, pedestrian network generation methods do not take into account the randomness of walking behaviors, cannot generate the off-road route segments and provide accurate route guidance. In this study, we proposed an approach for generating pedestrian network based on improved navigation graph. First, walking environment was regarded as the space with obstacles. The generation method of navigation points was improved to process the global concave vertices in polygon obstacles. Then, the optimization method of navigation points based on mergence area was proposed to effectively reduce the number of navigation points. Finally, the generation algorithm of navigation edges based on the dynamic cone-shaped sweep was proposed to generate sparse navigation edges. The experiment results showed that this approach can generate the improved navigation graph which data size can be largely reduced. The pedestrian network represented by the improved navigation graph can provide pedestrians with accurate route guidance. This approach can effectively solve the modelling problem of the pedestrian network in pedestrian navigation applications, having a certain practical value.