Navigation in Complex Space: An Bayesian Nash Equilibrium-Informed Agent-Based Model (Short Paper)

This study proposed an improved pedestrian evacuation ABM employing Bayesian Nash Equilibrium (BNE) to simulate more realistic and representative individual evacuating behaviours in complex scenarios. A set of vertical blockades with adjustable gate widths was introduced to establish a simulation space with narrow corridor and bottlenecks and to evaluate the influences of BNE on individual navigation in complex space. To better match with the evacuating behaviours in real-world scenarios, the decision-making criterion of BNE evacuees was improved to a multi-strategy combination, with 80% of evacuees taking the optimal strategy, 15% taking sub-optimal strategy, and 5% taking the third-best one. The preliminary results demonstrate a positive impact of BNE on individual navigation in complex space, showing a distinct decrease of evacuation time with increasing proportion of BNE evacuees. The non-monotonicity of the variations in evacuation time also indicates the dynamic adaptability of BNE in addressing immediate challenges (i.e. blockades and congestions), which identifies alternative and potential faster paths during evacuations. A detailed description of the proposed ABM and an analysis of relevant experimental results are provided in this paper. Several limitations are also identified.