Approaching humans in crowded and dynamic environments

This paper proposes an effective human approaching algorithm that enables a mobile service robot to safely and socially approach humans in dynamic and crowded environments. The proposed method is composed of two major functional blocks: (1) The approaching pose estimation (APE) is to estimate the approaching pose of both stationary and moving humans by combining the proposed extended personal space (EPS) of humans and environmental surrounding in the robot's vicinity; (2) The goal-oriented hybrid reciprocal velocity obstacles (G-HRVO) based motion planning by incorporating human states and the approaching pose into the conventional hybrid reciprocal velocity obstacles (HRVO) is to drive the robot towards the goal while proactively keeping a safe distance with humans and obstacles. The simulation results show that the our approach is capable of controlling the mobile robot to approach both stationary and moving humans in socially acceptable manners.