Etiquettes aware grasp motion planning for robotic handover operation

Abstract

For home-service robots one of the most common interactions with humans is the handing over objects using arms and hands of anthropomorphic robots. We propose advanced methods of determination of grasp sites for handover operations that incorporate especially etiquettes which factors include object shapes, object functions, and safety. We also address different operations for handover: one-handed handover with one grasp, two-step handover with midair re-grasp, and two-handed handover. We show the effectiveness of our algorithm for interaction between humans and robots with graphical simulations, and then it is tested with our humanoid home-service robot IDRO.