Evaluating geometric similarities between indoor navigation paths and navigation patterns of pedestrians through Geographic Information System and Multi-Criteria Decision Analysis

Abstract

The indoor paths conveyed to the end-users of navigation systems must align with their spatial reasoning to support their wayfinding process. To achieve this, the geometric properties of indoor paths followed by pedestrians should be similar to their navigation patterns. In this study, an integrated approach based on GIS and MCDA was proposed to assess geometric similarities between navigation networks and the navigation patterns of pedestrians (Figure A). Figure A. Flowchart of the proposed methodology Purpose: The mainpurpose of the study is to establish an integrated approach for determining geometric similarities between indoor navigation networks and the navigation patterns of pedestrians, supporting the wayfinding process in indoor navigation.Theory and Methods: In the proposed integrated approach, first, four commonly used navigation networks (MAT, Grid, MPRSSE, UCN) were generated using the indoor map of the study area. Next, an experiment was conducted to collect the navigation patterns of pedestrians. Third, various geometric similarity measures were computed for indoor paths and navigation patterns. Subsequently, the importance of each measure was determined using the Fuzzy AHP method. Utilizing the criteria weights, the TOPSIS method was employed to rank navigation networks based on their geometric similarity to the navigation patterns of pedestrians.Results: The results of the experimental study showed that the MPRSSE navigation network was the closest to the navigation patterns of pedestrians in terms of geometric similarity. The UCN navigation network was found to be the least similar, despite previous studies suggesting its suitability for length and turn minimization.Conclusion: In this study, an approach based on GIS and MCDA was proposed to assess the geometric similarity between navigation networks and the navigation patterns of pedestrians. Our experimental study suggests that medial axis extraction-based MPRSSE and MAT navigation networks are closer to navigation patterns in terms of geometric similarity. For future studies, additional criteria and combinations of navigation networks can be explored to achieve more comprehensive coverage and select a suitable navigation network for use in navigation systems.