Simulation of socially aware mobile robot navigation

Abstract

Traditional local planners often treat all pedestrians in the same way, ignoring differences in safety needs among age groups. To address this problem, this paper studies how age attributes can be used in socially-aware mobile robot navigation. The proposed framework includes two main parts: age-stage perception and differentiated social local planning. At the perception level, a YOLOv8n-based module is developed to detect and classify three target groups: children, adults, and the elderly. The experimental results show that this module can output target positions and age-stage categories in a stable way. At the planning level, this paper analyzes a local planning framework that combines a social force model with an environmental cost map. Simulation results show that this framework can generate collision-free trajectories with basic social avoidance behavior. Furthermore, an age-based parameter mapping strategy is proposed, allowing different age groups to correspond to different social force parameters. Although real-time closed-loop integration has not yet been achieved, this paper establishes a logical link between age-stage perception, parameter mapping, and local planning behavior.