Effect of prediction on the self-organization of pedestrian counter flow

Pedestrians may predict the behavior of others and then adjust their movement accordingly to avoid potential conflicts in advance. Motivated by this fact, we propose a predictive control theory-based pedestrian counter flow model, which describes the predictive mechanism underlying pedestrian self-organization phenomena. In this model, a pedestrian will make in-advance-avoid behavior based on the estimation of future moving gain within a given predictive length to reduce potential conflicts. The future gain in the present model is affected by three factors, i.e. the predictive length, the smooth degree of entrance and the influential area of coming pedestrians. Simulation results of the model show that increasing predictive length has a remarkable effect on reducing conflicts, improving pedestrian velocity, smoothing pedestrian movement and stabilizing the self-organized lanes. When enlarging the influential area of coming pedestrians, pedestrians tend to aggregate to the formed self-organized lanes, which makes the lanes wider and the lane number reduced. Interestingly, moderate enlargement (of the influential area) will reduce conflicts significantly, while excessive enlargement will lead to an increase in conflicts. We also discuss the predictive effect toward the smooth degree of entrance. When there are some formed self-organized lanes in the system, the effect is significant, and it will make the lanes more regular and stable, while when the existing lanes are unstable, the effect has little impact on the system.