Motion Planning of a Human-Aware Mobile Robot Merging into Dynamic Pedestrian Flow

摘要

Safe and efficient robot navigation for people and robots in dense human crowds is a challenging task because the close interaction between people with diverse behavioral intentions limits and complicates the movement space. Humans move safely and efficiently without disturbing movement space by recognizing and utilizing the flow which is formed by common self-organization phenomenon in crowds. Conventional systems have not clearly implemented algorithms for generating human-aware merging behavior. Furthermore, they are insufficient for robots to select the optimal pedestrian flow for merging and following, and to move safely and efficiently. We therefore propose a system that can join a social group without causing discomfort to pedestrians, and move safely and efficiently to a destination without disturbing the flow of the movement space. In simulation and experimental robot navigation, the proposed system generates human-aware behavior for improving safety and efficiency of navigation.