Human-Robot Harmonious Coexistence: A Robot Navigation Framework Follows Social Norms in Dynamic Scenarios

Abstract

Designing a path planner that provides humans with a comfortable experience is a crucial step in enabling robots to seamlessly integrate into human environments. In this paper, we propose a comprehensive framework that equips robots with the ability to navigate in human environments by taking into account social norms and predicting human trajectories. Firstly, we propose a generalized social space modeling method based on Gaussian Mixture Model (GMM). This model is used to constrain the robot's adherence to social norms and incorporates factors such as human posture, velocity, and group distribution. Secondly, to achieve collision-free navigation, we introduce a trajectory prediction method using a four-parameter logistic curve. This method considers human historical trajectory information, velocity constraints, and incorporates confidence weights. Finally, we use piecewise high-order polynomials to optimize robot's local trajectory spatiotemporally. The collision avoidance constraints between the robot and humans are carefully designed to maximize human comfort and robot sociability. To validate the effectiveness of our approach, we compare it with existing methods, and the results demonstrate a significant improvement in human comfort.