Hybrid-driven Trajectory Prediction Based on Group Emotion

TL;DR

This paper introduces a hybrid trajectory prediction method that combines data-driven and model-driven approaches, incorporating group emotion and social relations to improve accuracy and controllability in crowd movement modeling.

Contribution

It proposes a novel hybrid method integrating group emotion and social relations to enhance crowd trajectory prediction with better controllability and reduced data dependence.

Findings

Improved trajectory prediction accuracy in crowd simulations.

Effective incorporation of group emotion influences on individual movements.

Enhanced model controllability and generality.

Abstract

We present a hybrid-driven trajectory prediction method based on group emotion. The data driven and model driven methods are combined to make a compromise between the controllability, generality, and efficiency of the method on the basis of simulating more real crowd movements. A hybrid driven method is proposed to improve the reliability of the calculation results based on real crowd data, and ensure the controllability of the model. It reduces the dependence of our model on real data and realizes the complementary advantages of these two kinds of methods. In addition, we divide crowd into groups based on human relations in society. So our method can calculate the movements in different scales. We predict individual movement trajectories according to the trajectories of group and fully consider the influence of the group movement state on the individual movements. Besides we also propose a group emotion calculation method and our method also considers the effect of group emotion on crowd movements.