A Personalized Comfort Space With Variable Shape Based on Environmental Information for Robot Navigation in Homes

Abstract

When a robot navigates in home environments, its trajectories need not only to avoid collisions but also to meet the psychological comfort of family members. We propose a personalized comfort space with variable shape based on environmental information to enhance the comfort perception of different family members and ensure efficient navigation. Our method considers the different psychological demands of comfort in multiple directions based on the human pose and velocity. The personalized comfort distances for different family members inferred based on human characteristics serve as the range benchmarks of the comfort spaces. We also incorporate environmental information around the person to achieve shape adaptation with the obstacle distributions, particularly benefiting navigation in narrow areas. Additionally, we develop a costmap plugin based on robot operating system (ROS) for our comfort space, facilitating its integration with different navigation modules and enabling easy deployment onto robots. Experimental results demonstrate that our method improves human comfort while maintaining navigation success and efficiency.