Robot Motion and Grasping for Blindfold Handover

Abstract

Autonomous robots in human-robot interaction (HRI) recently are becoming part of human life as the number of services or personal robots increasingly used in our home. In order to fulfill the gap of HRI merit, we would like to propose a system of the autonomous robot motion and grasping creation for assisting the disabled person such as the blind people for handover tasks to help blind people in pick and place tasks. In this paper, we develop the robot motion to receive the object by handing over from the blindfold human to represent the blindness. To determine the target of the object and human hand, we implement the 6DOF pose detection using a point cloud and hand detection using the Single Shot Detection model in Deep learning for planning motion using 9DOF arm robot with hand. We finally experiment and evaluate the tasks from blindfold-robot handover tasks.