Construction of Decision Model for Determining the Behavior of Pedestrians Encountering at a L-shaped Corner by Considering Dynamics of Judgment Characteristics Caused by Interplays with the Other

Abstract

In the paper, we construct a decision model for determining the behavior of pedestrians encountering at a L-shaped corner taking into consideration the dynamics of judgment characteristics caused by interplays with each other. First, we formulate parameters that characterize the behavioral judgment of pedestrians according to the model, based on the game theory. Further, we extend the decision model to exhibit dynamic behaviors of changing their judgment characteristics using evolutionary game theory and reinforcement learning. Then, the extended model was experimented to verify the effectiveness by simulation. Based on such a model as to take human judgment characteristics into account, robot navigation control is expected to realize the safe operations even at indoor blind corners that contain a high risk of collision.