Intellectual Route Planning Methods for Realistic Agents' Movement

Abstract

Realistic pedestrian flow visualization is a complex problem of multi-agent modeling. It can be applied to different crowd dynamics solutions. This paper presents a new route planning method for agents that can be used in interactive simulation of crowd movement. It is based on game development technologies and modern artificial intelligence techniques. We use reinforcement learning to train the agent to navigate through moving pedestrians. A set of experiments were carried out and the results of the study were evaluated. The proposed approach is compared with existing methods and its perspectives are discussed.