From IFC BIM to Semantically Enriched 2.5D Indoor Navigation Graphs for Congestion-Aware Pedestrian Simulation
X. Wang (TU Delft - Architecture and the Built Environment)
G.A.K. Arroyo Ohori – Mentor (TU Delft - Architecture and the Built Environment)
B.M. Meijers – Mentor (TU Delft - Architecture and the Built Environment)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Indoor navigation in complex public buildings requires more than geometric shortest-path routing. In multi-level station environments, pedestrian movement is shaped by vertical connectors, access-control elements, semantic origin–destination regions, congestion, and heterogeneous walking behaviour. Although Industry Foundation Classes (IFC) Building Information Modeling (BIM) provides rich geometric and semantic information about building elements, it is not directly usable as a navigation or pedestrian simulation model.
This thesis develops and evaluates a reproducible workflow for transforming IFC BIM data into a semantically enriched 2.5D indoor navigation graph for congestion-aware pedestrian routing and simulation. The workflow interprets movement-relevant IFC semantics, extracts floor-based walkable layers, samples valid graph nodes, filters obstacles and restricted areas, and models vertical connectors and access-control elements as semantic graph components. The resulting graph is used as the movement substrate for graph-based routing and agent-based pedestrian simulation.
The workflow is first checked on a simplified two-floor pilot model and subsequently applied to Dadongmen Station of Hefei Metro Line 1. The final default station graph contains 18,454 nodes and 134,852 edges across three public movement levels. Six experiments evaluate static routing, congestion-aware replanning, pedestrian-profile effects, mixed-agent behaviour, algorithm-threshold sensitivity, and demand-load response.
The results show that congestion-aware replanning reduces waiting time and peak queue length, although it may increase mean travel time because agents use longer alternative routes. Elderly agents experience substantially higher travel and waiting burdens than normal agents. A sparse graph variant remains routable but produces larger queues and earlier demand-load degradation. The findings demonstrate that graph connectivity alone is insufficient: indoor navigation graphs should also be evaluated through movement-performance metrics. The results represent controlled methodological validation rather than calibrated operational passenger-flow prediction.